New_Validation_LightGBM.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "mergedDescs=pd.read_csv(\"merged_Descs_sorted.csv\")\n",
    "data=mergedDescs.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_names=pd.read_csv(\"ExternalID.csv\", header=None)\n",
    "train_names=pd.read_csv(\"TrainID.csv\", header=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "data[\"BCF\"] = data.apply(lambda row: 0 if row[\"logBCF\"] < 3.0 else 1, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = data.drop(['logBCF'], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0    798\n",
       "1    209\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# for splitting the dataset, selected rows based on unique values in the 'Name' column using or (|)\n",
    "# unique values were chosen from original datasets provided by authors of publication\n",
    "test= data[(data['Name']==729)|\n",
    "           (data['Name']==367)|\n",
    "           (data['Name']==475)|\n",
    "           (data['Name']==951)|\n",
    "           (data['Name']==21)|\n",
    "           (data['Name']==810)|\n",
    "           (data['Name']==1009)|\n",
    "           (data['Name']==273)|\n",
    "           (data['Name']==168)|\n",
    "           (data['Name']==586)|\n",
    "           (data['Name']==594)|\n",
    "           (data['Name']==22)|\n",
    "           (data['Name']==13)|\n",
    "           (data['Name']==941)|\n",
    "           (data['Name']==4)|\n",
    "           (data['Name']==253)|\n",
    "           (data['Name']==748)|\n",
    "           (data['Name']==807)|\n",
    "           (data['Name']==23)|\n",
    "           (data['Name']==953)|\n",
    "           (data['Name']==876)|\n",
    "           (data['Name']==571)|\n",
    "           (data['Name']==600)|\n",
    "           (data['Name']==372)|\n",
    "           (data['Name']==492)|\n",
    "           (data['Name']==802)|\n",
    "           (data['Name']==3)|\n",
    "           (data['Name']==118)|\n",
    "           (data['Name']==606)|\n",
    "           (data['Name']==42)|\n",
    "           (data['Name']==569)|\n",
    "           (data['Name']==505)|\n",
    "           (data['Name']==324)|\n",
    "           (data['Name']==588)|\n",
    "           (data['Name']==173)|\n",
    "           (data['Name']==65)|\n",
    "           (data['Name']==867)|\n",
    "           (data['Name']==145)|\n",
    "           (data['Name']==1032)|\n",
    "           (data['Name']==98)|\n",
    "           (data['Name']==101)|\n",
    "           (data['Name']==478)|\n",
    "           (data['Name']==30)|\n",
    "           (data['Name']==959)|\n",
    "           (data['Name']==960)|\n",
    "           (data['Name']==396)|\n",
    "           (data['Name']==621)|\n",
    "           (data['Name']==860)|\n",
    "           (data['Name']==559)|\n",
    "           (data['Name']==100)|\n",
    "           (data['Name']==568)|\n",
    "           (data['Name']==847)|\n",
    "           (data['Name']==194)|\n",
    "           (data['Name']==120)|\n",
    "           (data['Name']==24)|\n",
    "           (data['Name']==970)|\n",
    "           (data['Name']==149)|\n",
    "           (data['Name']==99)|\n",
    "           (data['Name']==204)|\n",
    "           (data['Name']==183)|\n",
    "           (data['Name']==10)|\n",
    "           (data['Name']==652)|\n",
    "           (data['Name']==840)|\n",
    "           (data['Name']==774)|\n",
    "           (data['Name']==576)|\n",
    "           (data['Name']==496)|\n",
    "           (data['Name']==74)|\n",
    "           (data['Name']==557)|\n",
    "           (data['Name']==871)|\n",
    "           (data['Name']==1030)|\n",
    "           (data['Name']==518)|\n",
    "           (data['Name']==126)|\n",
    "           (data['Name']==964)|\n",
    "           (data['Name']==215)|\n",
    "           (data['Name']==73)|\n",
    "           (data['Name']==555)|\n",
    "           (data['Name']==412)|\n",
    "           (data['Name']==150)|\n",
    "           (data['Name']==719)|\n",
    "           (data['Name']==537)|\n",
    "           (data['Name']==612)|\n",
    "           (data['Name']==136)|\n",
    "           (data['Name']==75)|\n",
    "           (data['Name']==26)|\n",
    "           (data['Name']==707)|\n",
    "           (data['Name']==424)|\n",
    "           (data['Name']==917)|\n",
    "           (data['Name']==565)|\n",
    "           (data['Name']==79)|\n",
    "           (data['Name']==66)|\n",
    "           (data['Name']==248)|\n",
    "           (data['Name']==155)|\n",
    "           (data['Name']==237)|\n",
    "           (data['Name']==721)|\n",
    "           (data['Name']==212)|\n",
    "           (data['Name']==523)|\n",
    "           (data['Name']==541)|\n",
    "           (data['Name']==112)|\n",
    "           (data['Name']==668)|\n",
    "           (data['Name']==222)|\n",
    "           (data['Name']==940)|\n",
    "           (data['Name']==491)|\n",
    "           (data['Name']==924)|\n",
    "           (data['Name']==146)|\n",
    "           (data['Name']==64)|\n",
    "           (data['Name']==539)|\n",
    "           (data['Name']==190)|\n",
    "           (data['Name']==1007)|\n",
    "           (data['Name']==511)|\n",
    "           (data['Name']==854)|\n",
    "           (data['Name']==334)|\n",
    "           (data['Name']==617)|\n",
    "           (data['Name']==409)|\n",
    "           (data['Name']==582)|\n",
    "           (data['Name']==61)|\n",
    "           (data['Name']==156)|\n",
    "           (data['Name']==1036)|\n",
    "           (data['Name']==543)|\n",
    "           (data['Name']==706)|\n",
    "           (data['Name']==95)|\n",
    "           (data['Name']==442)|\n",
    "           (data['Name']==649)|\n",
    "           (data['Name']==989)|\n",
    "           (data['Name']==121)|\n",
    "           (data['Name']==388)|\n",
    "           (data['Name']==708)|\n",
    "           (data['Name']==616)|\n",
    "           (data['Name']==436)|\n",
    "           (data['Name']==444)|\n",
    "           (data['Name']==968)|\n",
    "           (data['Name']==734)|\n",
    "           (data['Name']==205)|\n",
    "           (data['Name']==611)|\n",
    "           (data['Name']==610)|\n",
    "           (data['Name']==92)|\n",
    "           (data['Name']==819)|\n",
    "           (data['Name']==535)|\n",
    "           (data['Name']==653)|\n",
    "           (data['Name']==1033)|\n",
    "           (data['Name']==943)|\n",
    "           (data['Name']==656)|\n",
    "           (data['Name']==288)|\n",
    "           (data['Name']==347)|\n",
    "           (data['Name']==698)|\n",
    "           (data['Name']==227)|\n",
    "           (data['Name']==337)|\n",
    "           (data['Name']==745)|\n",
    "           (data['Name']==293)|\n",
    "           (data['Name']==452)|\n",
    "           (data['Name']==609)|\n",
    "           (data['Name']==384)|\n",
    "           (data['Name']==763)|\n",
    "           (data['Name']==363)|\n",
    "           (data['Name']==368)|\n",
    "           (data['Name']==1006)|\n",
    "           (data['Name']==546)|\n",
    "           (data['Name']==137)|\n",
    "           (data['Name']==797)|\n",
    "           (data['Name']==762)|\n",
    "           (data['Name']==1018)|\n",
    "           (data['Name']==153)|\n",
    "           (data['Name']==769)|\n",
    "           (data['Name']==312)|\n",
    "           (data['Name']==1020)|\n",
    "           (data['Name']==216)|\n",
    "           (data['Name']==151)|\n",
    "           (data['Name']==815)|\n",
    "           (data['Name']==800)|\n",
    "           (data['Name']==290)|\n",
    "           (data['Name']==345)|\n",
    "           (data['Name']==371)|\n",
    "           (data['Name']==401)|\n",
    "           (data['Name']==733)|\n",
    "           (data['Name']==768)|\n",
    "           (data['Name']==40)|\n",
    "           (data['Name']==58)|\n",
    "           (data['Name']==948)|\n",
    "           (data['Name']==132)|\n",
    "           (data['Name']==728)|\n",
    "           (data['Name']==676)|\n",
    "           (data['Name']==855)|\n",
    "           (data['Name']==325)|\n",
    "           (data['Name']==737)|\n",
    "           (data['Name']==309)|\n",
    "           (data['Name']==355)|\n",
    "           (data['Name']==925)|\n",
    "           (data['Name']==8)|\n",
    "           (data['Name']==743)|\n",
    "           (data['Name']==913)|\n",
    "           (data['Name']==782)|\n",
    "           (data['Name']==428)|\n",
    "           (data['Name']==301)|\n",
    "           (data['Name']==431)|\n",
    "           (data['Name']==592)|\n",
    "           (data['Name']==15)|\n",
    "           (data['Name']==903)|\n",
    "           (data['Name']==276)|\n",
    "           (data['Name']==315)|\n",
    "           (data['Name']==302)|\n",
    "           (data['Name']==525)|\n",
    "           (data['Name']==316)|\n",
    "           (data['Name']==35)|\n",
    "           (data['Name']==422)|\n",
    "           (data['Name']==723)|\n",
    "           (data['Name']==59)|\n",
    "           (data['Name']==1021)|\n",
    "           (data['Name']==184)|\n",
    "           (data['Name']==53)|\n",
    "           (data['Name']==892)|\n",
    "           (data['Name']==69)|\n",
    "           (data['Name']==973)|\n",
    "           (data['Name']==322)|\n",
    "           (data['Name']==310)|\n",
    "           (data['Name']==623)|\n",
    "           (data['Name']==965)|\n",
    "           (data['Name']==124)|\n",
    "           (data['Name']==420)|\n",
    "           (data['Name']==16)|\n",
    "           (data['Name']==882)|\n",
    "           (data['Name']==961)|\n",
    "           (data['Name']==90)|\n",
    "           (data['Name']==283)|\n",
    "           (data['Name']==839)|\n",
    "           (data['Name']==39)|\n",
    "           (data['Name']==856)|\n",
    "           (data['Name']==752)|\n",
    "           (data['Name']==1035)|\n",
    "           (data['Name']==540)|\n",
    "           (data['Name']==1004)|\n",
    "           (data['Name']==545)|\n",
    "           (data['Name']==863)|\n",
    "           (data['Name']==296)|\n",
    "           (data['Name']==447)|\n",
    "           (data['Name']==558)|\n",
    "           (data['Name']==286)|\n",
    "           (data['Name']==348)|\n",
    "           (data['Name']==111)|\n",
    "           (data['Name']==831)|\n",
    "           (data['Name']==130)|\n",
    "           (data['Name']==710)|\n",
    "           (data['Name']==56)|\n",
    "           (data['Name']==242)|\n",
    "           (data['Name']==779)|\n",
    "           (data['Name']==643)|\n",
    "           (data['Name']==574)|\n",
    "           (data['Name']==672)|\n",
    "           (data['Name']==356)|\n",
    "           (data['Name']==2)|\n",
    "           (data['Name']==147)|\n",
    "           (data['Name']==317)|\n",
    "           (data['Name']==502)|\n",
    "           (data['Name']==853)|\n",
    "           (data['Name']==932)|\n",
    "           (data['Name']==937)|\n",
    "           (data['Name']==437)|\n",
    "           (data['Name']==143)|\n",
    "           (data['Name']==922)|\n",
    "           (data['Name']==773)|\n",
    "           (data['Name']==659)|\n",
    "           (data['Name']==378)|\n",
    "           (data['Name']==230)|\n",
    "           (data['Name']==241)|\n",
    "           (data['Name']==603)|\n",
    "           (data['Name']==343)|\n",
    "           (data['Name']==524)|\n",
    "           (data['Name']==28)|\n",
    "           (data['Name']==307)|\n",
    "           (data['Name']==259)|\n",
    "           (data['Name']==342)|\n",
    "           (data['Name']==454)|\n",
    "           (data['Name']==791)|\n",
    "           (data['Name']==81)|\n",
    "           (data['Name']==697)|\n",
    "           (data['Name']==209)|\n",
    "           (data['Name']==415)|\n",
    "           (data['Name']==125)|\n",
    "           (data['Name']==599)|\n",
    "           (data['Name']==662)|\n",
    "           (data['Name']==593)|\n",
    "           (data['Name']==313)|\n",
    "           (data['Name']==628)|\n",
    "           (data['Name']==664)|\n",
    "           (data['Name']==705)|\n",
    "           (data['Name']==117)|\n",
    "           (data['Name']==760)|\n",
    "           (data['Name']==878)|\n",
    "           (data['Name']==534)|\n",
    "           (data['Name']==232)|\n",
    "           (data['Name']==234)|\n",
    "           (data['Name']==977)|\n",
    "           (data['Name']==318)|\n",
    "           (data['Name']==688)|\n",
    "           (data['Name']==104)|\n",
    "           (data['Name']==969)|\n",
    "           (data['Name']==292)|\n",
    "           (data['Name']==445)|\n",
    "           (data['Name']==482)|\n",
    "           (data['Name']==678)|\n",
    "           (data['Name']==618)|\n",
    "           (data['Name']==266)|\n",
    "           (data['Name']==642)|\n",
    "           (data['Name']==140)|\n",
    "           (data['Name']==249)|\n",
    "           (data['Name']==873)|\n",
    "           (data['Name']==984)|\n",
    "           (data['Name']==421)|\n",
    "           (data['Name']==279)|\n",
    "           (data['Name']==185)|\n",
    "           (data['Name']==758)|\n",
    "           (data['Name']==201)|\n",
    "           (data['Name']==682)|\n",
    "           (data['Name']==211)|\n",
    "           (data['Name']==1029)|\n",
    "           (data['Name']==981)|\n",
    "           (data['Name']==887)|\n",
    "           (data['Name']==133)|\n",
    "           (data['Name']==391)|\n",
    "           (data['Name']==684)|\n",
    "           (data['Name']==670)|\n",
    "           (data['Name']==694)|\n",
    "           (data['Name']==641)|\n",
    "           (data['Name']==247)|\n",
    "           (data['Name']==256)|\n",
    "           (data['Name']==686)|\n",
    "           (data['Name']==218)|\n",
    "           (data['Name']==893)|\n",
    "           (data['Name']==875)|\n",
    "           (data['Name']==18)|\n",
    "           (data['Name']==679)|\n",
    "           (data['Name']==280)|\n",
    "           (data['Name']==999)|\n",
    "           (data['Name']==323)|\n",
    "           (data['Name']==687)|\n",
    "           (data['Name']==683)\n",
    "          ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# the same approach as for the test set above\n",
    "train=data[(data['Name']==485)|\n",
    "           (data['Name']==381)|\n",
    "        (data['Name']==282)|\n",
    "        (data['Name']==333)|\n",
    "        (data['Name']==275)|\n",
    "        (data['Name']==494)|\n",
    "        (data['Name']==161)|\n",
    "        (data['Name']==584)|\n",
    "        (data['Name']==823)|\n",
    "        (data['Name']==410)|\n",
    "        (data['Name']==352)|\n",
    "        (data['Name']==908)|\n",
    "        (data['Name']==261)|\n",
    "        (data['Name']==715)|\n",
    "        (data['Name']==551)|\n",
    "        (data['Name']==597)|\n",
    "        (data['Name']==711)|\n",
    "        (data['Name']==716)|\n",
    "        (data['Name']==846)|\n",
    "        (data['Name']==103)|\n",
    "        (data['Name']==224)|\n",
    "        (data['Name']==54)|\n",
    "        (data['Name']==792)|\n",
    "        (data['Name']==1)|\n",
    "        (data['Name']==457)|\n",
    "        (data['Name']==730)|\n",
    "        (data['Name']==829)|\n",
    "        (data['Name']==869)|\n",
    "        (data['Name']==219)|\n",
    "        (data['Name']==601)|\n",
    "        (data['Name']==105)|\n",
    "        (data['Name']==37)|\n",
    "        (data['Name']==148)|\n",
    "        (data['Name']==532)|\n",
    "        (data['Name']==416)|\n",
    "        (data['Name']==509)|\n",
    "        (data['Name']==453)|\n",
    "        (data['Name']==468)|\n",
    "        (data['Name']==179)|\n",
    "        (data['Name']==154)|\n",
    "        (data['Name']==615)|\n",
    "        (data['Name']==771)|\n",
    "        (data['Name']==14)|\n",
    "        (data['Name']==195)|\n",
    "        (data['Name']==265)|\n",
    "        (data['Name']==640)|\n",
    "        (data['Name']==129)|\n",
    "        (data['Name']==335)|\n",
    "        (data['Name']==455)|\n",
    "        (data['Name']==477)|\n",
    "        (data['Name']==498)|\n",
    "        (data['Name']==573)|\n",
    "        (data['Name']==1025)|\n",
    "        (data['Name']==199)|\n",
    "        (data['Name']==134)|\n",
    "        (data['Name']==91)|\n",
    "        (data['Name']==268)|\n",
    "        (data['Name']==197)|\n",
    "        (data['Name']==515)|\n",
    "        (data['Name']==80)|\n",
    "        (data['Name']==508)|\n",
    "        (data['Name']==838)|\n",
    "        (data['Name']==461)|\n",
    "        (data['Name']==722)|\n",
    "        (data['Name']==203)|\n",
    "        (data['Name']==604)|\n",
    "        (data['Name']==41)|\n",
    "        (data['Name']==11)|\n",
    "        (data['Name']==48)|\n",
    "        (data['Name']==872)|\n",
    "        (data['Name']==198)|\n",
    "        (data['Name']==735)|\n",
    "        (data['Name']==229)|\n",
    "        (data['Name']==906)|\n",
    "        (data['Name']==390)|\n",
    "        (data['Name']==581)|\n",
    "        (data['Name']==598)|\n",
    "        (data['Name']==1031)|\n",
    "        (data['Name']==221)|\n",
    "        (data['Name']==107)|\n",
    "        (data['Name']==57)|\n",
    "        (data['Name']==258)|\n",
    "        (data['Name']==171)|\n",
    "        (data['Name']==693)|\n",
    "        (data['Name']==645)|\n",
    "        (data['Name']==31)|\n",
    "        (data['Name']==884)|\n",
    "        (data['Name']==210)|\n",
    "        (data['Name']==563)|\n",
    "        (data['Name']==650)|\n",
    "        (data['Name']==361)|\n",
    "        (data['Name']==364)|\n",
    "        (data['Name']==828)|\n",
    "        (data['Name']==12)|\n",
    "        (data['Name']==564)|\n",
    "        (data['Name']==174)|\n",
    "        (data['Name']==536)|\n",
    "        (data['Name']==97)|\n",
    "        (data['Name']==519)|\n",
    "        (data['Name']==483)|\n",
    "        (data['Name']==963)|\n",
    "        (data['Name']==277)|\n",
    "        (data['Name']==583)|\n",
    "        (data['Name']==552)|\n",
    "        (data['Name']==514)|\n",
    "        (data['Name']==119)|\n",
    "        (data['Name']==304)|\n",
    "        (data['Name']==362)|\n",
    "        (data['Name']==213)|\n",
    "        (data['Name']==71)|\n",
    "        (data['Name']==231)|\n",
    "        (data['Name']==51)|\n",
    "        (data['Name']==942)|\n",
    "        (data['Name']==1019)|\n",
    "        (data['Name']==113)|\n",
    "        (data['Name']==660)|\n",
    "        (data['Name']==510)|\n",
    "        (data['Name']==142)|\n",
    "        (data['Name']==47)|\n",
    "        (data['Name']==966)|\n",
    "        (data['Name']==985)|\n",
    "        (data['Name']==157)|\n",
    "        (data['Name']==17)|\n",
    "        (data['Name']==46)|\n",
    "        (data['Name']==799)|\n",
    "        (data['Name']==544)|\n",
    "        (data['Name']==139)|\n",
    "        (data['Name']==400)|\n",
    "        (data['Name']==62)|\n",
    "        (data['Name']==252)|\n",
    "        (data['Name']==786)|\n",
    "        (data['Name']==820)|\n",
    "        (data['Name']==251)|\n",
    "        (data['Name']==862)|\n",
    "        (data['Name']==935)|\n",
    "        (data['Name']==671)|\n",
    "        (data['Name']==624)|\n",
    "        (data['Name']==633)|\n",
    "        (data['Name']==978)|\n",
    "        (data['Name']==383)|\n",
    "        (data['Name']==516)|\n",
    "        (data['Name']==891)|\n",
    "        (data['Name']==931)|\n",
    "        (data['Name']==178)|\n",
    "        (data['Name']==202)|\n",
    "        (data['Name']==562)|\n",
    "        (data['Name']==625)|\n",
    "        (data['Name']==60)|\n",
    "        (data['Name']==96)|\n",
    "        (data['Name']==920)|\n",
    "        (data['Name']==547)|\n",
    "        (data['Name']==351)|\n",
    "        (data['Name']==635)|\n",
    "        (data['Name']==776)|\n",
    "        (data['Name']==228)|\n",
    "        (data['Name']==45)|\n",
    "        (data['Name']==88)|\n",
    "        (data['Name']==192)|\n",
    "        (data['Name']==44)|\n",
    "        (data['Name']==549)|\n",
    "        (data['Name']==70)|\n",
    "        (data['Name']==499)|\n",
    "        (data['Name']==67)|\n",
    "        (data['Name']==696)|\n",
    "        (data['Name']==602)|\n",
    "        (data['Name']==36)|\n",
    "        (data['Name']==952)|\n",
    "        (data['Name']==975)|\n",
    "        (data['Name']==848)|\n",
    "        (data['Name']==440)|\n",
    "        (data['Name']==1005)|\n",
    "        (data['Name']==94)|\n",
    "        (data['Name']==550)|\n",
    "        (data['Name']==837)|\n",
    "        (data['Name']==1015)|\n",
    "        (data['Name']==507)|\n",
    "        (data['Name']==780)|\n",
    "        (data['Name']==34)|\n",
    "        (data['Name']==527)|\n",
    "        (data['Name']==407)|\n",
    "        (data['Name']==858)|\n",
    "        (data['Name']==577)|\n",
    "        (data['Name']==956)|\n",
    "        (data['Name']==667)|\n",
    "        (data['Name']==1022)|\n",
    "        (data['Name']==162)|\n",
    "        (data['Name']==620)|\n",
    "        (data['Name']==651)|\n",
    "        (data['Name']==830)|\n",
    "        (data['Name']==513)|\n",
    "        (data['Name']==865)|\n",
    "        (data['Name']==188)|\n",
    "        (data['Name']==298)|\n",
    "        (data['Name']==626)|\n",
    "        (data['Name']==548)|\n",
    "        (data['Name']==303)|\n",
    "        (data['Name']==84)|\n",
    "        (data['Name']==947)|\n",
    "        (data['Name']==566)|\n",
    "        (data['Name']==954)|\n",
    "        (data['Name']==506)|\n",
    "        (data['Name']==267)|\n",
    "        (data['Name']==319)|\n",
    "        (data['Name']==353)|\n",
    "        (data['Name']==929)|\n",
    "        (data['Name']==83)|\n",
    "        (data['Name']==833)|\n",
    "        (data['Name']==9)|\n",
    "        (data['Name']==967)|\n",
    "        (data['Name']==880)|\n",
    "        (data['Name']==957)|\n",
    "        (data['Name']==27)|\n",
    "        (data['Name']==822)|\n",
    "        (data['Name']==979)|\n",
    "        (data['Name']==332)|\n",
    "        (data['Name']==1003)|\n",
    "        (data['Name']==556)|\n",
    "        (data['Name']==843)|\n",
    "        (data['Name']==888)|\n",
    "        (data['Name']==5)|\n",
    "        (data['Name']==533)|\n",
    "        (data['Name']==330)|\n",
    "        (data['Name']==781)|\n",
    "        (data['Name']==927)|\n",
    "        (data['Name']==526)|\n",
    "        (data['Name']==25)|\n",
    "        (data['Name']==114)|\n",
    "        (data['Name']==945)|\n",
    "        (data['Name']==834)|\n",
    "        (data['Name']==764)|\n",
    "        (data['Name']==859)|\n",
    "        (data['Name']==223)|\n",
    "        (data['Name']==375)|\n",
    "        (data['Name']==1010)|\n",
    "        (data['Name']==832)|\n",
    "        (data['Name']==425)|\n",
    "        (data['Name']==634)|\n",
    "        (data['Name']==704)|\n",
    "        (data['Name']==844)|\n",
    "        (data['Name']==87)|\n",
    "        (data['Name']==877)|\n",
    "        (data['Name']==20)|\n",
    "        (data['Name']==852)|\n",
    "        (data['Name']==1001)|\n",
    "        (data['Name']==938)|\n",
    "        (data['Name']==495)|\n",
    "        (data['Name']==665)|\n",
    "        (data['Name']==427)|\n",
    "        (data['Name']==845)|\n",
    "        (data['Name']==135)|\n",
    "        (data['Name']==585)|\n",
    "        (data['Name']==365)|\n",
    "        (data['Name']==918)|\n",
    "        (data['Name']==187)|\n",
    "        (data['Name']==169)|\n",
    "        (data['Name']==474)|\n",
    "        (data['Name']==619)|\n",
    "        (data['Name']==939)|\n",
    "        (data['Name']==567)|\n",
    "        (data['Name']==666)|\n",
    "        (data['Name']==864)|\n",
    "        (data['Name']==486)|\n",
    "        (data['Name']==914)|\n",
    "        (data['Name']==622)|\n",
    "        (data['Name']==849)|\n",
    "        (data['Name']==836)|\n",
    "        (data['Name']==614)|\n",
    "        (data['Name']==467)|\n",
    "        (data['Name']==340)|\n",
    "        (data['Name']==759)|\n",
    "        (data['Name']==481)|\n",
    "        (data['Name']==374)|\n",
    "        (data['Name']==1008)|\n",
    "        (data['Name']==77)|\n",
    "        (data['Name']==958)|\n",
    "        (data['Name']==189)|\n",
    "        (data['Name']==528)|\n",
    "        (data['Name']==1012)|\n",
    "        (data['Name']==226)|\n",
    "        (data['Name']==466)|\n",
    "        (data['Name']==1002)|\n",
    "        (data['Name']==419)|\n",
    "        (data['Name']==257)|\n",
    "        (data['Name']==271)|\n",
    "        (data['Name']==373)|\n",
    "        (data['Name']==393)|\n",
    "        (data['Name']==191)|\n",
    "        (data['Name']==572)|\n",
    "        (data['Name']==49)|\n",
    "        (data['Name']==287)|\n",
    "        (data['Name']==857)|\n",
    "        (data['Name']==472)|\n",
    "        (data['Name']==811)|\n",
    "        (data['Name']==790)|\n",
    "        (data['Name']==850)|\n",
    "        (data['Name']==934)|\n",
    "        (data['Name']==661)|\n",
    "        (data['Name']==784)|\n",
    "        (data['Name']==632)|\n",
    "        (data['Name']==955)|\n",
    "        (data['Name']==200)|\n",
    "        (data['Name']==55)|\n",
    "        (data['Name']==736)|\n",
    "        (data['Name']==928)|\n",
    "        (data['Name']==68)|\n",
    "        (data['Name']==43)|\n",
    "        (data['Name']==500)|\n",
    "        (data['Name']==1034)|\n",
    "        (data['Name']==180)|\n",
    "        (data['Name']==439)|\n",
    "        (data['Name']==497)|\n",
    "        (data['Name']==264)|\n",
    "        (data['Name']==450)|\n",
    "        (data['Name']==269)|\n",
    "        (data['Name']==731)|\n",
    "        (data['Name']==911)|\n",
    "        (data['Name']==403)|\n",
    "        (data['Name']==542)|\n",
    "        (data['Name']==1013)|\n",
    "        (data['Name']==673)|\n",
    "        (data['Name']==115)|\n",
    "        (data['Name']==561)|\n",
    "        (data['Name']==122)|\n",
    "        (data['Name']==331)|\n",
    "        (data['Name']==578)|\n",
    "        (data['Name']==405)|\n",
    "        (data['Name']==714)|\n",
    "        (data['Name']==812)|\n",
    "        (data['Name']==742)|\n",
    "        (data['Name']==423)|\n",
    "        (data['Name']==756)|\n",
    "        (data['Name']==321)|\n",
    "        (data['Name']==746)|\n",
    "        (data['Name']==208)|\n",
    "        (data['Name']==738)|\n",
    "        (data['Name']==278)|\n",
    "        (data['Name']==285)|\n",
    "        (data['Name']==306)|\n",
    "        (data['Name']==327)|\n",
    "        (data['Name']==350)|\n",
    "        (data['Name']==357)|\n",
    "        (data['Name']==379)|\n",
    "        (data['Name']==399)|\n",
    "        (data['Name']==433)|\n",
    "        (data['Name']==448)|\n",
    "        (data['Name']==744)|\n",
    "        (data['Name']==750)|\n",
    "        (data['Name']==821)|\n",
    "        (data['Name']==1016)|\n",
    "        (data['Name']==910)|\n",
    "        (data['Name']==933)|\n",
    "        (data['Name']==596)|\n",
    "        (data['Name']==675)|\n",
    "        (data['Name']==490)|\n",
    "        (data['Name']==349)|\n",
    "        (data['Name']==1028)|\n",
    "        (data['Name']==554)|\n",
    "        (data['Name']==712)|\n",
    "        (data['Name']==504)|\n",
    "        (data['Name']==813)|\n",
    "        (data['Name']==1024)|\n",
    "        (data['Name']==976)|\n",
    "        (data['Name']==72)|\n",
    "        (data['Name']==426)|\n",
    "        (data['Name']==108)|\n",
    "        (data['Name']==255)|\n",
    "        (data['Name']==326)|\n",
    "        (data['Name']==339)|\n",
    "        (data['Name']==449)|\n",
    "        (data['Name']==469)|\n",
    "        (data['Name']==971)|\n",
    "        (data['Name']==1017)|\n",
    "        (data['Name']==946)|\n",
    "        (data['Name']==575)|\n",
    "        (data['Name']==131)|\n",
    "        (data['Name']==487)|\n",
    "        (data['Name']==438)|\n",
    "        (data['Name']==669)|\n",
    "        (data['Name']==700)|\n",
    "        (data['Name']==329)|\n",
    "        (data['Name']==775)|\n",
    "        (data['Name']==842)|\n",
    "        (data['Name']==522)|\n",
    "        (data['Name']==972)|\n",
    "        (data['Name']==1014)|\n",
    "        (data['Name']==883)|\n",
    "        (data['Name']==912)|\n",
    "        (data['Name']==233)|\n",
    "        (data['Name']==717)|\n",
    "        (data['Name']==186)|\n",
    "        (data['Name']==32)|\n",
    "        (data['Name']==629)|\n",
    "        (data['Name']==370)|\n",
    "        (data['Name']==489)|\n",
    "        (data['Name']==354)|\n",
    "        (data['Name']==727)|\n",
    "        (data['Name']==291)|\n",
    "        (data['Name']==1026)|\n",
    "        (data['Name']==464)|\n",
    "        (data['Name']==236)|\n",
    "        (data['Name']==369)|\n",
    "        (data['Name']==754)|\n",
    "        (data['Name']==789)|\n",
    "        (data['Name']==962)|\n",
    "        (data['Name']==785)|\n",
    "        (data['Name']==926)|\n",
    "        (data['Name']==553)|\n",
    "        (data['Name']==359)|\n",
    "        (data['Name']==206)|\n",
    "        (data['Name']==827)|\n",
    "        (data['Name']==909)|\n",
    "        (data['Name']==320)|\n",
    "        (data['Name']==560)|\n",
    "        (data['Name']==116)|\n",
    "        (data['Name']==709)|\n",
    "        (data['Name']==414)|\n",
    "        (data['Name']==644)|\n",
    "        (data['Name']==605)|\n",
    "        (data['Name']==907)|\n",
    "        (data['Name']==806)|\n",
    "        (data['Name']==591)|\n",
    "        (data['Name']==479)|\n",
    "        (data['Name']==608)|\n",
    "        (data['Name']==788)|\n",
    "        (data['Name']==50)|\n",
    "        (data['Name']==757)|\n",
    "        (data['Name']==93)|\n",
    "        (data['Name']==109)|\n",
    "        (data['Name']==377)|\n",
    "        (data['Name']==127)|\n",
    "        (data['Name']==366)|\n",
    "        (data['Name']==861)|\n",
    "        (data['Name']==681)|\n",
    "        (data['Name']==801)|\n",
    "        (data['Name']==851)|\n",
    "        (data['Name']==930)|\n",
    "        (data['Name']==803)|\n",
    "        (data['Name']==76)|\n",
    "        (data['Name']==870)|\n",
    "        (data['Name']==63)|\n",
    "        (data['Name']==33)|\n",
    "        (data['Name']==777)|\n",
    "        (data['Name']==986)|\n",
    "        (data['Name']==1023)|\n",
    "        (data['Name']==284)|\n",
    "        (data['Name']==718)|\n",
    "        (data['Name']==724)|\n",
    "        (data['Name']==295)|\n",
    "        (data['Name']==783)|\n",
    "        (data['Name']==905)|\n",
    "        (data['Name']==476)|\n",
    "        (data['Name']==772)|\n",
    "        (data['Name']==809)|\n",
    "        (data['Name']==835)|\n",
    "        (data['Name']==164)|\n",
    "        (data['Name']==761)|\n",
    "        (data['Name']==825)|\n",
    "        (data['Name']==235)|\n",
    "        (data['Name']==488)|\n",
    "        (data['Name']==82)|\n",
    "        (data['Name']==473)|\n",
    "        (data['Name']==392)|\n",
    "        (data['Name']==896)|\n",
    "        (data['Name']==281)|\n",
    "        (data['Name']==299)|\n",
    "        (data['Name']==7)|\n",
    "        (data['Name']==607)|\n",
    "        (data['Name']==658)|\n",
    "        (data['Name']==382)|\n",
    "        (data['Name']==314)|\n",
    "        (data['Name']==170)|\n",
    "        (data['Name']==794)|\n",
    "        (data['Name']==720)|\n",
    "        (data['Name']==1011)|\n",
    "        (data['Name']==459)|\n",
    "        (data['Name']==990)|\n",
    "        (data['Name']==879)|\n",
    "        (data['Name']==732)|\n",
    "        (data['Name']==78)|\n",
    "        (data['Name']==787)|\n",
    "        (data['Name']==86)|\n",
    "        (data['Name']==916)|\n",
    "        (data['Name']==176)|\n",
    "        (data['Name']==297)|\n",
    "        (data['Name']==766)|\n",
    "        (data['Name']==531)|\n",
    "        (data['Name']==796)|\n",
    "        (data['Name']==1027)|\n",
    "        (data['Name']==982)|\n",
    "        (data['Name']==866)|\n",
    "        (data['Name']==336)|\n",
    "        (data['Name']==165)|\n",
    "        (data['Name']==740)|\n",
    "        (data['Name']==239)|\n",
    "        (data['Name']==646)|\n",
    "        (data['Name']==85)|\n",
    "        (data['Name']==755)|\n",
    "        (data['Name']==451)|\n",
    "        (data['Name']==220)|\n",
    "        (data['Name']==713)|\n",
    "        (data['Name']==418)|\n",
    "        (data['Name']==983)|\n",
    "        (data['Name']==380)|\n",
    "        (data['Name']==590)|\n",
    "        (data['Name']==432)|\n",
    "        (data['Name']==570)|\n",
    "        (data['Name']==463)|\n",
    "        (data['Name']==595)|\n",
    "        (data['Name']==529)|\n",
    "        (data['Name']==770)|\n",
    "        (data['Name']==395)|\n",
    "        (data['Name']==895)|\n",
    "        (data['Name']==214)|\n",
    "        (data['Name']==89)|\n",
    "        (data['Name']==793)|\n",
    "        (data['Name']==739)|\n",
    "        (data['Name']==741)|\n",
    "        (data['Name']==753)|\n",
    "        (data['Name']==460)|\n",
    "        (data['Name']==874)|\n",
    "        (data['Name']==484)|\n",
    "        (data['Name']==358)|\n",
    "        (data['Name']==29)|\n",
    "        (data['Name']==638)|\n",
    "        (data['Name']==346)|\n",
    "        (data['Name']==881)|\n",
    "        (data['Name']==376)|\n",
    "        (data['Name']==446)|\n",
    "        (data['Name']==38)|\n",
    "        (data['Name']==995)|\n",
    "        (data['Name']==949)|\n",
    "        (data['Name']==919)|\n",
    "        (data['Name']==167)|\n",
    "        (data['Name']==805)|\n",
    "        (data['Name']==980)|\n",
    "        (data['Name']==458)|\n",
    "        (data['Name']==580)|\n",
    "        (data['Name']==394)|\n",
    "        (data['Name']==443)|\n",
    "        (data['Name']==398)|\n",
    "        (data['Name']==886)|\n",
    "        (data['Name']==996)|\n",
    "        (data['Name']==465)|\n",
    "        (data['Name']==244)|\n",
    "        (data['Name']==703)|\n",
    "        (data['Name']==841)|\n",
    "        (data['Name']==138)|\n",
    "        (data['Name']==207)|\n",
    "        (data['Name']==19)|\n",
    "        (data['Name']==480)|\n",
    "        (data['Name']==411)|\n",
    "        (data['Name']==102)|\n",
    "        (data['Name']==663)|\n",
    "        (data['Name']==904)|\n",
    "        (data['Name']==767)|\n",
    "        (data['Name']==579)|\n",
    "        (data['Name']==974)|\n",
    "        (data['Name']==587)|\n",
    "        (data['Name']==344)|\n",
    "        (data['Name']==123)|\n",
    "        (data['Name']==441)|\n",
    "        (data['Name']==402)|\n",
    "        (data['Name']==814)|\n",
    "        (data['Name']==462)|\n",
    "        (data['Name']==921)|\n",
    "        (data['Name']==385)|\n",
    "        (data['Name']==778)|\n",
    "        (data['Name']==950)|\n",
    "        (data['Name']==655)|\n",
    "        (data['Name']==701)|\n",
    "        (data['Name']==795)|\n",
    "        (data['Name']==816)|\n",
    "        (data['Name']==181)|\n",
    "        (data['Name']==901)|\n",
    "        (data['Name']==936)|\n",
    "        (data['Name']==456)|\n",
    "        (data['Name']==885)|\n",
    "        (data['Name']==196)|\n",
    "        (data['Name']==294)|\n",
    "        (data['Name']==141)|\n",
    "        (data['Name']==408)|\n",
    "        (data['Name']==900)|\n",
    "        (data['Name']==630)|\n",
    "        (data['Name']==1000)|\n",
    "        (data['Name']==387)|\n",
    "        (data['Name']==152)|\n",
    "        (data['Name']==413)|\n",
    "        (data['Name']==915)|\n",
    "        (data['Name']==699)|\n",
    "        (data['Name']==890)|\n",
    "        (data['Name']==818)|\n",
    "        (data['Name']==128)|\n",
    "        (data['Name']==328)|\n",
    "        (data['Name']==270)|\n",
    "        (data['Name']==308)|\n",
    "        (data['Name']==749)|\n",
    "        (data['Name']==804)|\n",
    "        (data['Name']==631)|\n",
    "        (data['Name']==674)|\n",
    "        (data['Name']==889)|\n",
    "        (data['Name']==589)|\n",
    "        (data['Name']==144)|\n",
    "        (data['Name']==225)|\n",
    "        (data['Name']==158)|\n",
    "        (data['Name']==289)|\n",
    "        (data['Name']==654)|\n",
    "        (data['Name']==826)|\n",
    "        (data['Name']==272)|\n",
    "        (data['Name']==52)|\n",
    "        (data['Name']==637)|\n",
    "        (data['Name']==691)|\n",
    "        (data['Name']==689)|\n",
    "        (data['Name']==520)|\n",
    "        (data['Name']==657)|\n",
    "        (data['Name']==471)|\n",
    "        (data['Name']==166)|\n",
    "        (data['Name']==300)|\n",
    "        (data['Name']==765)|\n",
    "        (data['Name']==987)|\n",
    "        (data['Name']==639)|\n",
    "        (data['Name']==944)|\n",
    "        (data['Name']==994)|\n",
    "        (data['Name']==274)|\n",
    "        (data['Name']==254)|\n",
    "        (data['Name']==159)|\n",
    "        (data['Name']==240)|\n",
    "        (data['Name']==106)|\n",
    "        (data['Name']==530)|\n",
    "        (data['Name']==868)|\n",
    "        (data['Name']==429)|\n",
    "        (data['Name']==260)|\n",
    "        (data['Name']==417)|\n",
    "        (data['Name']==389)|\n",
    "        (data['Name']==690)|\n",
    "        (data['Name']==193)|\n",
    "        (data['Name']==685)|\n",
    "        (data['Name']==538)|\n",
    "        (data['Name']==406)|\n",
    "        (data['Name']==217)|\n",
    "        (data['Name']==648)|\n",
    "        (data['Name']==680)|\n",
    "        (data['Name']==798)|\n",
    "        (data['Name']==898)|\n",
    "        (data['Name']==677)|\n",
    "        (data['Name']==360)|\n",
    "        (data['Name']==163)|\n",
    "        (data['Name']==110)|\n",
    "        (data['Name']==386)|\n",
    "        (data['Name']==899)|\n",
    "        (data['Name']==991)|\n",
    "        (data['Name']==923)|\n",
    "        (data['Name']==692)|\n",
    "        (data['Name']==246)|\n",
    "        (data['Name']==647)|\n",
    "        (data['Name']==992)|\n",
    "        (data['Name']==172)|\n",
    "        (data['Name']==250)|\n",
    "        (data['Name']==894)|\n",
    "        (data['Name']==434)|\n",
    "        (data['Name']==182)|\n",
    "        (data['Name']==897)|\n",
    "        (data['Name']==993)|\n",
    "        (data['Name']==245)|\n",
    "        (data['Name']==751)|\n",
    "        (data['Name']==243)|\n",
    "        (data['Name']==997)|\n",
    "        (data['Name']==998)|\n",
    "        (data['Name']==988)|\n",
    "        (data['Name']==430)|\n",
    "        (data['Name']==435)|\n",
    "        (data['Name']==747)|\n",
    "        (data['Name']==305)|\n",
    "        (data['Name']==512)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "#count = 0\n",
    "test1 = pd.DataFrame()\n",
    "train1 = pd.DataFrame()\n",
    "#for index, row in train.iterrows():\n",
    " #   count += 1\n",
    "  #  if (count%3)==0:\n",
    "   #     test1 = test1.append(row)\n",
    "    #else:\n",
    "     #   train1 = train1.append(row)\n",
    "train1 = train\n",
    "test1 = test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 673 entries, 0 to 1006\n",
      "Columns: 553 entries, Name to BCF\n",
      "dtypes: float64(375), int64(178)\n",
      "memory usage: 2.8 MB\n"
     ]
    }
   ],
   "source": [
    "train1.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 334 entries, 2 to 1004\n",
      "Columns: 553 entries, Name to BCF\n",
      "dtypes: float64(375), int64(178)\n",
      "memory usage: 1.4 MB\n"
     ]
    }
   ],
   "source": [
    "test1.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "train = train[[\"AMR\", \"VP-3\", \"SsOH\", \"BCF\"]]\n",
    "test = test[[\"AMR\", \"VP-3\", \"SsOH\", \"BCF\"]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = train1.drop(['BCF'], axis=1)\n",
    "y_train = train1['BCF']\n",
    "X_test = test1.drop(['BCF'], axis=1)\n",
    "y_test = test1['BCF']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0    532\n",
      "1    141\n",
      "Name: BCF, dtype: int64\n",
      "0    266\n",
      "1     68\n",
      "Name: BCF, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# just to see how it looks like..\n",
    "print(y_train.value_counts())\n",
    "print(y_test.value_counts())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1. SVM (Linear)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1.1. Imbalanced model towards nB compounds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.svm import SVC\n",
    "svclassifier = SVC(kernel='linear').fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[240  26]\n",
      " [ 23  45]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.91      0.90      0.91       266\n",
      "           1       0.63      0.66      0.65        68\n",
      "\n",
      "    accuracy                           0.85       334\n",
      "   macro avg       0.77      0.78      0.78       334\n",
      "weighted avg       0.86      0.85      0.85       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import classification_report, confusion_matrix\n",
    "\n",
    "y_pred = svclassifier.predict(X_test)\n",
    "\n",
    "print(confusion_matrix(y_test,y_pred))\n",
    "print(classification_report(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this is the new result\n",
    "M1_sp_SVM_linear=0.90\n",
    "M1_se_SVM_linear=0.66\n",
    "M1_acc_SVM_linear=0.85\n",
    "# this is the allowed error:\n",
    "M1_err_sp=0.00\n",
    "M1_err_se=0.0194\n",
    "M1_err_acc=0.00\n",
    "# this is the reference result:\n",
    "M1_sp=0.91\n",
    "M1_se=0.7206\n",
    "M1_acc=0.87\n",
    "# this is the gain or loss: new result minus reference result\n",
    "M1_sp_SVM_linear_eval=(M1_sp_SVM_linear-M1_sp)\n",
    "M1_se_SVM_linear_eval=(M1_se_SVM_linear-M1_se)\n",
    "M1_acc_SVM_linear_eval=(M1_acc_SVM_linear-M1_acc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1.2.1. Balanced Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0    532\n",
       "1    141\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# BEFORE\n",
    "train1.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0    141\n",
       "1    141\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "positive_cases = train1[train1['BCF'] == 1]\n",
    "negative_sample = train1[train1['BCF'] == 0].sample(141, random_state=0)\n",
    "train1_balanced = pd.concat([positive_cases,negative_sample])\n",
    "train1_balanced.sort_index(inplace=True)\n",
    "\n",
    "X_train_bal = train1_balanced.drop(['BCF'], axis=1)\n",
    "y_train_bal = train1_balanced['BCF']\n",
    "\n",
    "# AFTER\n",
    "train1_balanced.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.svm import SVC\n",
    "svclassifier = SVC(kernel='linear').fit(X_train_bal, y_train_bal)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[199  67]\n",
      " [ 10  58]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.95      0.75      0.84       266\n",
      "           1       0.46      0.85      0.60        68\n",
      "\n",
      "    accuracy                           0.77       334\n",
      "   macro avg       0.71      0.80      0.72       334\n",
      "weighted avg       0.85      0.77      0.79       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "y_pred = svclassifier.predict(X_test)\n",
    "\n",
    "print(confusion_matrix(y_test,y_pred))\n",
    "print(classification_report(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this is the new result\n",
    "M2_sp_SVM_linear=0.75\n",
    "M2_se_SVM_linear=0.85\n",
    "M2_acc_SVM_linear=0.77\n",
    "# this is the allowed error:\n",
    "M2_err_sp=0.00\n",
    "M2_err_se=0.0129\n",
    "M2_err_acc=0.01\n",
    "# this is the reference result:\n",
    "M2_sp=0.83\n",
    "M2_se=0.91\n",
    "M2_acc=0.85\n",
    "# this is the gain or loss: new result minus reference result\n",
    "M2_sp_SVM_linear_eval=(M2_sp_SVM_linear-M2_sp)\n",
    "M2_se_SVM_linear_eval=(M2_se_SVM_linear-M2_se)\n",
    "M2_acc_SVM_linear_eval=(M2_acc_SVM_linear-M2_acc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1.2.2 Balanced Model-SMOTE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "import imblearn\n",
    "from imblearn.over_sampling import SMOTE\n",
    "smote = SMOTE(random_state=0)\n",
    "X_train_resampled, y_train_resampled = smote.fit_sample(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "svclassifier = SVC(kernel='linear').fit(X_train_resampled, y_train_resampled)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[233  33]\n",
      " [ 17  51]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.93      0.88      0.90       266\n",
      "           1       0.61      0.75      0.67        68\n",
      "\n",
      "    accuracy                           0.85       334\n",
      "   macro avg       0.77      0.81      0.79       334\n",
      "weighted avg       0.87      0.85      0.86       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "y_pred = svclassifier.predict(X_test)\n",
    "\n",
    "print(confusion_matrix(y_test,y_pred))\n",
    "print(classification_report(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this is the new result\n",
    "M2_2_sp_SVM_linear=0.88\n",
    "M2_2_se_SVM_linear=0.75\n",
    "M2_2_acc_SVM_linear=0.85\n",
    "# this is the allowed error:\n",
    "M2_2_err_sp=0.06\n",
    "M2_2_err_se=0.0571\n",
    "M2_2_err_acc=0.04\n",
    "# this is the reference result:\n",
    "M2_2_sp=0.89\n",
    "M2_2_se=0.84\n",
    "M2_2_acc=0.88\n",
    "# this is the gain or loss: new result minus reference result\n",
    "M2_2_sp_SVM_linear_eval=(M2_2_sp_SVM_linear-M2_2_sp)\n",
    "M2_2_se_SVM_linear_eval=(M2_2_se_SVM_linear-M2_2_se)\n",
    "M2_2_acc_SVM_linear_eval=(M2_2_acc_SVM_linear-M2_2_acc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1.3. Imbalanced model towards B compounds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0    532\n",
       "1    141\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# BEFORE\n",
    "train1.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0     28\n",
       "1    107\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "positive_cases = train1[train1['BCF'] == 1].sample(107, random_state=0)\n",
    "negative_sample = train1[train1['BCF'] == 0].sample(28, random_state=0)\n",
    "train1_imbalanced = pd.concat([positive_cases,negative_sample])\n",
    "train1_imbalanced.sort_index(inplace=True)\n",
    "\n",
    "X_train_imbal = train1_imbalanced.drop(['BCF'], axis=1)\n",
    "y_train_imbal = train1_imbalanced['BCF']\n",
    "# AFTER\n",
    "train1_imbalanced.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "svclassifier = SVC(kernel='linear', C=7.0).fit(X_train_imbal, y_train_imbal)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[161 105]\n",
      " [  8  60]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.95      0.61      0.74       266\n",
      "           1       0.36      0.88      0.52        68\n",
      "\n",
      "    accuracy                           0.66       334\n",
      "   macro avg       0.66      0.74      0.63       334\n",
      "weighted avg       0.83      0.66      0.69       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "y_pred = svclassifier.predict(X_test)\n",
    "\n",
    "print(confusion_matrix(y_test,y_pred))\n",
    "print(classification_report(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this is the new result\n",
    "M3_sp_SVM_linear=0.61\n",
    "M3_se_SVM_linear=0.88\n",
    "M3_acc_SVM_linear=0.66\n",
    "# this is the allowed error:\n",
    "M3_err_sp=0.0692\n",
    "M3_err_se=0.00\n",
    "M3_err_acc=0.057\n",
    "# this is the reference result:\n",
    "M3_sp=0.6\n",
    "M3_se=1.00\n",
    "M3_acc=0.68\n",
    "# this is the gain or loss: new result minus reference result\n",
    "M3_sp_SVM_linear_eval=(M3_sp_SVM_linear-M3_sp)\n",
    "M3_se_SVM_linear_eval=(M3_se_SVM_linear-M3_se)\n",
    "M3_acc_SVM_linear_eval=(M3_acc_SVM_linear-M3_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Model</th>\n",
       "      <th>Sp_new_res</th>\n",
       "      <th>Sp_allowed_err</th>\n",
       "      <th>Sp_ref_result</th>\n",
       "      <th>Sp_gain/loss</th>\n",
       "      <th>Se_new_res</th>\n",
       "      <th>Se_allowed_err</th>\n",
       "      <th>Se_ref_result</th>\n",
       "      <th>Se_gain/loss</th>\n",
       "      <th>Acc_new_res</th>\n",
       "      <th>Acc_allowed_err</th>\n",
       "      <th>Acc_ref_result</th>\n",
       "      <th>Acc_gain/loss</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>M1:SVM_linear</td>\n",
       "      <td>0.90</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.91</td>\n",
       "      <td>-0.01</td>\n",
       "      <td>0.66</td>\n",
       "      <td>0.0194</td>\n",
       "      <td>0.7206</td>\n",
       "      <td>-0.0606</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.87</td>\n",
       "      <td>-0.02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>M2:SVM_linear</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.83</td>\n",
       "      <td>-0.08</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.0129</td>\n",
       "      <td>0.9100</td>\n",
       "      <td>-0.0600</td>\n",
       "      <td>0.77</td>\n",
       "      <td>0.010</td>\n",
       "      <td>0.85</td>\n",
       "      <td>-0.08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>M2_2:SVM_linear</td>\n",
       "      <td>0.88</td>\n",
       "      <td>0.0600</td>\n",
       "      <td>0.89</td>\n",
       "      <td>-0.01</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.0571</td>\n",
       "      <td>0.8400</td>\n",
       "      <td>-0.0900</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.040</td>\n",
       "      <td>0.88</td>\n",
       "      <td>-0.03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>M3:SVM_linear</td>\n",
       "      <td>0.61</td>\n",
       "      <td>0.0692</td>\n",
       "      <td>0.60</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.88</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>1.0000</td>\n",
       "      <td>-0.1200</td>\n",
       "      <td>0.66</td>\n",
       "      <td>0.057</td>\n",
       "      <td>0.68</td>\n",
       "      <td>-0.02</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             Model  Sp_new_res  Sp_allowed_err  Sp_ref_result  Sp_gain/loss  \\\n",
       "0    M1:SVM_linear        0.90          0.0000           0.91         -0.01   \n",
       "1    M2:SVM_linear        0.75          0.0000           0.83         -0.08   \n",
       "2  M2_2:SVM_linear        0.88          0.0600           0.89         -0.01   \n",
       "3    M3:SVM_linear        0.61          0.0692           0.60          0.01   \n",
       "\n",
       "   Se_new_res  Se_allowed_err  Se_ref_result  Se_gain/loss  Acc_new_res  \\\n",
       "0        0.66          0.0194         0.7206       -0.0606         0.85   \n",
       "1        0.85          0.0129         0.9100       -0.0600         0.77   \n",
       "2        0.75          0.0571         0.8400       -0.0900         0.85   \n",
       "3        0.88          0.0000         1.0000       -0.1200         0.66   \n",
       "\n",
       "   Acc_allowed_err  Acc_ref_result  Acc_gain/loss  \n",
       "0            0.000            0.87          -0.02  \n",
       "1            0.010            0.85          -0.08  \n",
       "2            0.040            0.88          -0.03  \n",
       "3            0.057            0.68          -0.02  "
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results = pd.DataFrame({ 'Model' : pd.Categorical([\"M1:SVM_linear\",\"M2:SVM_linear\",\"M2_2:SVM_linear\",\"M3:SVM_linear\"]),\n",
    "                    'Sp_new_res' : [M1_sp_SVM_linear,M2_sp_SVM_linear,M2_2_sp_SVM_linear,M3_sp_SVM_linear],\n",
    "                    'Sp_allowed_err' : [M1_err_sp,M2_err_sp,M2_2_err_sp,M3_err_sp],\n",
    "                    'Sp_ref_result' : [M1_sp,M2_sp,M2_2_sp,M3_sp],\n",
    "                    'Sp_gain/loss' : [M1_sp_SVM_linear_eval,M2_sp_SVM_linear_eval,M2_2_sp_SVM_linear_eval,M3_sp_SVM_linear_eval],\n",
    "                    'Se_new_res' : [M1_se_SVM_linear,M2_se_SVM_linear,M2_2_se_SVM_linear,M3_se_SVM_linear],\n",
    "                    'Se_allowed_err' : [M1_err_se,M2_err_se,M2_2_err_se,M3_err_se],\n",
    "                    'Se_ref_result' : [M1_se,M2_se,M2_2_se,M3_se],\n",
    "                    'Se_gain/loss' : [M1_se_SVM_linear_eval,M2_se_SVM_linear_eval,M2_2_se_SVM_linear_eval,M3_se_SVM_linear_eval],\n",
    "                    'Acc_new_res' : [M1_acc_SVM_linear,M2_acc_SVM_linear,M2_2_acc_SVM_linear,M3_acc_SVM_linear],\n",
    "                    'Acc_allowed_err' : [M1_err_acc,M2_err_acc,M2_2_err_acc,M3_err_acc],\n",
    "                    'Acc_ref_result' : [M1_acc,M2_acc,M2_2_acc,M3_acc],\n",
    "                    'Acc_gain/loss' : [M1_acc_SVM_linear_eval,M2_acc_SVM_linear_eval,M2_2_acc_SVM_linear_eval,M3_acc_SVM_linear_eval], })\n",
    "results\n",
    "## it is more important to find a model with higher sensitivity and \n",
    "## reasonable specificity for the classification of BCF!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2. Naive Bayers (Gaussian Kernel)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2.1. Imbalanced model towards nB compounds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "#results.drop(results.tail(4).index,inplace=True)\n",
    "#results.drop()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[205  61]\n",
      " [ 12  56]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.94      0.77      0.85       266\n",
      "           1       0.48      0.82      0.61        68\n",
      "\n",
      "    accuracy                           0.78       334\n",
      "   macro avg       0.71      0.80      0.73       334\n",
      "weighted avg       0.85      0.78      0.80       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.naive_bayes import GaussianNB \n",
    "from sklearn.metrics import classification_report, confusion_matrix\n",
    "gnb = GaussianNB().fit(X_train, y_train)\n",
    "y_pred = gnb.predict(X_test) \n",
    "print(confusion_matrix(y_test,y_pred))\n",
    "print(classification_report(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this is the new result\n",
    "M1_sp_naive_gaussian=0.77\n",
    "M1_se_naive_gaussian=0.82\n",
    "M1_acc_naive_gaussian=0.78\n",
    "# this is the allowed error:\n",
    "M1_err_sp=0.00\n",
    "M1_err_se=0.0194\n",
    "M1_err_acc=0.00\n",
    "# this is the reference result:\n",
    "M1_sp=0.91\n",
    "M1_se=0.7206\n",
    "M1_acc=0.87\n",
    "# this is the gain or loss: new result minus reference result\n",
    "M1_sp_naive_gaussian_eval=(M1_sp_naive_gaussian-M1_sp)\n",
    "M1_se_naive_gaussian_eval=(M1_se_naive_gaussian-M1_se)\n",
    "M1_acc_naive_gaussian_eval=(M1_acc_naive_gaussian-M1_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "naive_gaussian_m1 = {'Model':\"M1:Naive_gaussian\", 'Sp_new_res':M1_sp_naive_gaussian, 'Sp_allowed_err':M1_err_sp, 'Sp_ref_result':M1_sp, 'Sp_gain/loss':M1_sp_naive_gaussian_eval, 'Se_new_res':M1_se_naive_gaussian, 'Se_allowed_err':M1_err_se, 'Se_ref_result':M1_se, 'Se_gain/loss':M1_se_naive_gaussian_eval, 'Acc_new_res':M1_acc_naive_gaussian, 'Acc_allowed_err':M1_err_acc, 'Acc_ref_result':M1_acc, 'Acc_gain/loss':M1_acc_naive_gaussian_eval}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = results.append(naive_gaussian_m1, ignore_index=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2.2.1. Balanced Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0    532\n",
       "1    141\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train1.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0    141\n",
       "1    141\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "positive_cases = train1[train1['BCF'] == 1]\n",
    "negative_sample = train1[train1['BCF'] == 0].sample(141, random_state=0)\n",
    "train1_balanced = pd.concat([positive_cases,negative_sample])\n",
    "train1_balanced.sort_index(inplace=True)\n",
    "\n",
    "X_train_bal = train1_balanced.drop(['BCF'], axis=1)\n",
    "y_train_bal = train1_balanced['BCF']\n",
    "\n",
    "# AFTER\n",
    "train1_balanced.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[201  65]\n",
      " [  9  59]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.96      0.76      0.84       266\n",
      "           1       0.48      0.87      0.61        68\n",
      "\n",
      "    accuracy                           0.78       334\n",
      "   macro avg       0.72      0.81      0.73       334\n",
      "weighted avg       0.86      0.78      0.80       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.naive_bayes import GaussianNB \n",
    "from sklearn.metrics import classification_report, confusion_matrix\n",
    "gnb = GaussianNB().fit(X_train_bal, y_train_bal)\n",
    "y_pred = gnb.predict(X_test) \n",
    "print(confusion_matrix(y_test,y_pred))\n",
    "print(classification_report(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this is the new result\n",
    "M2_sp_naive_gaussian=0.76\n",
    "M2_se_naive_gaussian=0.87\n",
    "M2_acc_naive_gaussian=0.78\n",
    "# this is the allowed error:\n",
    "M2_err_sp=0.00\n",
    "M2_err_se=0.0129\n",
    "M2_err_acc=0.01\n",
    "# this is the reference result:\n",
    "M2_sp=0.83\n",
    "M2_se=0.91\n",
    "M2_acc=0.85\n",
    "# this is the gain or loss: new result minus reference result\n",
    "M2_sp_naive_gaussian_eval=(M2_sp_naive_gaussian-M2_sp)\n",
    "M2_se_naive_gaussian_eval=(M2_se_naive_gaussian-M2_se)\n",
    "M2_acc_naive_gaussian_eval=(M2_acc_naive_gaussian-M2_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "naive_gaussian_m2 = {'Model':\"M2:Naive_gaussian\", 'Sp_new_res':M2_sp_naive_gaussian, 'Sp_allowed_err':M2_err_sp, 'Sp_ref_result':M2_sp, 'Sp_gain/loss':M2_sp_naive_gaussian_eval, 'Se_new_res':M2_se_naive_gaussian, 'Se_allowed_err':M2_err_se, 'Se_ref_result':M2_se, 'Se_gain/loss':M2_se_naive_gaussian_eval, 'Acc_new_res':M2_acc_naive_gaussian, 'Acc_allowed_err':M2_err_acc, 'Acc_ref_result':M2_acc, 'Acc_gain/loss':M2_acc_naive_gaussian_eval}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = results.append(naive_gaussian_m2, ignore_index=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2.2.2 Balanced Model-SMOTE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "import imblearn\n",
    "from imblearn.over_sampling import SMOTE\n",
    "smote = SMOTE(random_state=0)\n",
    "X_train_resampled, y_train_resampled = smote.fit_sample(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[196  70]\n",
      " [ 10  58]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.95      0.74      0.83       266\n",
      "           1       0.45      0.85      0.59        68\n",
      "\n",
      "    accuracy                           0.76       334\n",
      "   macro avg       0.70      0.79      0.71       334\n",
      "weighted avg       0.85      0.76      0.78       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.naive_bayes import GaussianNB \n",
    "from sklearn.metrics import classification_report, confusion_matrix\n",
    "gnb = GaussianNB().fit(X_train_resampled, y_train_resampled)\n",
    "y_pred = gnb.predict(X_test) \n",
    "print(confusion_matrix(y_test,y_pred))\n",
    "print(classification_report(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this is the new result\n",
    "M2_2_sp_naive_gaussian=0.74\n",
    "M2_2_se_naive_gaussian=0.85\n",
    "M2_2_acc_naive_gaussian=0.76\n",
    "# this is the allowed error:\n",
    "M2_2_err_sp=0.06\n",
    "M2_2_err_se=0.0571\n",
    "M2_2_err_acc=0.04\n",
    "# this is the reference result:\n",
    "M2_2_sp=0.89\n",
    "M2_2_se=0.84\n",
    "M2_2_acc=0.88\n",
    "# this is the gain or loss: new result minus reference result\n",
    "M2_2_sp_naive_gaussian_eval=(M2_2_sp_naive_gaussian-M2_2_sp)\n",
    "M2_2_se_naive_gaussian_eval=(M2_2_se_naive_gaussian-M2_2_se)\n",
    "M2_2_acc_naive_gaussian_eval=(M2_2_acc_naive_gaussian-M2_2_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "naive_gaussian_m2_2 = {'Model':\"M2_2:Naive_gaussian\", 'Sp_new_res':M2_2_sp_naive_gaussian, 'Sp_allowed_err':M2_2_err_sp, 'Sp_ref_result':M2_2_sp, 'Sp_gain/loss':M2_2_sp_naive_gaussian_eval, 'Se_new_res':M2_2_se_naive_gaussian, 'Se_allowed_err':M2_2_err_se, 'Se_ref_result':M2_2_se, 'Se_gain/loss':M2_2_se_naive_gaussian_eval, 'Acc_new_res':M2_2_acc_naive_gaussian, 'Acc_allowed_err':M2_2_err_acc, 'Acc_ref_result':M2_2_acc, 'Acc_gain/loss':M2_2_acc_naive_gaussian_eval}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = results.append(naive_gaussian_m2_2, ignore_index=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "#results.drop(results.tail(5).index,inplace=True)\n",
    "#results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2.3. Imbalanced model towards B compounds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0    532\n",
       "1    141\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train1.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0     28\n",
       "1    107\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "positive_cases = train1[train1['BCF'] == 1].sample(107, random_state=0)\n",
    "negative_sample = train1[train1['BCF'] == 0].sample(28, random_state=0)\n",
    "train1_imbalanced = pd.concat([positive_cases,negative_sample])\n",
    "train1_imbalanced.sort_index(inplace=True)\n",
    "\n",
    "X_train_imbal = train1_imbalanced.drop(['BCF'], axis=1)\n",
    "y_train_imbal = train1_imbalanced['BCF']\n",
    "# AFTER\n",
    "train1_imbalanced.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[209  57]\n",
      " [ 17  51]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.92      0.79      0.85       266\n",
      "           1       0.47      0.75      0.58        68\n",
      "\n",
      "    accuracy                           0.78       334\n",
      "   macro avg       0.70      0.77      0.71       334\n",
      "weighted avg       0.83      0.78      0.79       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.naive_bayes import GaussianNB \n",
    "from sklearn.metrics import classification_report, confusion_matrix\n",
    "gnb = GaussianNB().fit(X_train_imbal, y_train_imbal)\n",
    "y_pred = gnb.predict(X_test) \n",
    "print(confusion_matrix(y_test,y_pred))\n",
    "print(classification_report(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this is the new result\n",
    "M3_sp_naive_gaussian=0.79\n",
    "M3_se_naive_gaussian=0.75\n",
    "M3_acc_naive_gaussian=0.78\n",
    "# this is the allowed error:\n",
    "M3_err_sp=0.0692\n",
    "M3_err_se=0.00\n",
    "M3_err_acc=0.057\n",
    "# this is the reference result:\n",
    "M3_sp=0.6\n",
    "M3_se=1.00\n",
    "M3_acc=0.68\n",
    "# this is the gain or loss: new result minus reference result\n",
    "M3_sp_naive_gaussian_eval=(M3_sp_naive_gaussian-M3_sp)\n",
    "M3_se_naive_gaussian_eval=(M3_se_naive_gaussian-M3_se)\n",
    "M3_acc_naive_gaussian_eval=(M3_acc_naive_gaussian-M3_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "naive_gaussian_m3 = {'Model':\"M3:Naive_gaussian\", 'Sp_new_res':M3_sp_naive_gaussian, 'Sp_allowed_err':M3_err_sp, 'Sp_ref_result':M3_sp, 'Sp_gain/loss':M3_sp_naive_gaussian_eval, 'Se_new_res':M3_se_naive_gaussian, 'Se_allowed_err':M3_err_se, 'Se_ref_result':M3_se, 'Se_gain/loss':M3_se_naive_gaussian_eval, 'Acc_new_res':M3_acc_naive_gaussian, 'Acc_allowed_err':M3_err_acc, 'Acc_ref_result':M3_acc, 'Acc_gain/loss':M3_acc_naive_gaussian_eval}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Model</th>\n",
       "      <th>Sp_new_res</th>\n",
       "      <th>Sp_allowed_err</th>\n",
       "      <th>Sp_ref_result</th>\n",
       "      <th>Sp_gain/loss</th>\n",
       "      <th>Se_new_res</th>\n",
       "      <th>Se_allowed_err</th>\n",
       "      <th>Se_ref_result</th>\n",
       "      <th>Se_gain/loss</th>\n",
       "      <th>Acc_new_res</th>\n",
       "      <th>Acc_allowed_err</th>\n",
       "      <th>Acc_ref_result</th>\n",
       "      <th>Acc_gain/loss</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>M1:SVM_linear</td>\n",
       "      <td>0.90</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.91</td>\n",
       "      <td>-0.01</td>\n",
       "      <td>0.66</td>\n",
       "      <td>0.0194</td>\n",
       "      <td>0.7206</td>\n",
       "      <td>-0.0606</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.87</td>\n",
       "      <td>-0.02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>M2:SVM_linear</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.83</td>\n",
       "      <td>-0.08</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.0129</td>\n",
       "      <td>0.9100</td>\n",
       "      <td>-0.0600</td>\n",
       "      <td>0.77</td>\n",
       "      <td>0.010</td>\n",
       "      <td>0.85</td>\n",
       "      <td>-0.08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>M2_2:SVM_linear</td>\n",
       "      <td>0.88</td>\n",
       "      <td>0.0600</td>\n",
       "      <td>0.89</td>\n",
       "      <td>-0.01</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.0571</td>\n",
       "      <td>0.8400</td>\n",
       "      <td>-0.0900</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.040</td>\n",
       "      <td>0.88</td>\n",
       "      <td>-0.03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>M3:SVM_linear</td>\n",
       "      <td>0.61</td>\n",
       "      <td>0.0692</td>\n",
       "      <td>0.60</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.88</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>1.0000</td>\n",
       "      <td>-0.1200</td>\n",
       "      <td>0.66</td>\n",
       "      <td>0.057</td>\n",
       "      <td>0.68</td>\n",
       "      <td>-0.02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>M1:Naive_gaussian</td>\n",
       "      <td>0.77</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.91</td>\n",
       "      <td>-0.14</td>\n",
       "      <td>0.82</td>\n",
       "      <td>0.0194</td>\n",
       "      <td>0.7206</td>\n",
       "      <td>0.0994</td>\n",
       "      <td>0.78</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.87</td>\n",
       "      <td>-0.09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>M2:Naive_gaussian</td>\n",
       "      <td>0.76</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.83</td>\n",
       "      <td>-0.07</td>\n",
       "      <td>0.87</td>\n",
       "      <td>0.0129</td>\n",
       "      <td>0.9100</td>\n",
       "      <td>-0.0400</td>\n",
       "      <td>0.78</td>\n",
       "      <td>0.010</td>\n",
       "      <td>0.85</td>\n",
       "      <td>-0.07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>M2_2:Naive_gaussian</td>\n",
       "      <td>0.74</td>\n",
       "      <td>0.0600</td>\n",
       "      <td>0.89</td>\n",
       "      <td>-0.15</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.0571</td>\n",
       "      <td>0.8400</td>\n",
       "      <td>0.0100</td>\n",
       "      <td>0.76</td>\n",
       "      <td>0.040</td>\n",
       "      <td>0.88</td>\n",
       "      <td>-0.12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>M3:Naive_gaussian</td>\n",
       "      <td>0.79</td>\n",
       "      <td>0.0692</td>\n",
       "      <td>0.60</td>\n",
       "      <td>0.19</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>1.0000</td>\n",
       "      <td>-0.2500</td>\n",
       "      <td>0.78</td>\n",
       "      <td>0.057</td>\n",
       "      <td>0.68</td>\n",
       "      <td>0.10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Model  Sp_new_res  Sp_allowed_err  Sp_ref_result  \\\n",
       "0        M1:SVM_linear        0.90          0.0000           0.91   \n",
       "1        M2:SVM_linear        0.75          0.0000           0.83   \n",
       "2      M2_2:SVM_linear        0.88          0.0600           0.89   \n",
       "3        M3:SVM_linear        0.61          0.0692           0.60   \n",
       "4    M1:Naive_gaussian        0.77          0.0000           0.91   \n",
       "5    M2:Naive_gaussian        0.76          0.0000           0.83   \n",
       "6  M2_2:Naive_gaussian        0.74          0.0600           0.89   \n",
       "7    M3:Naive_gaussian        0.79          0.0692           0.60   \n",
       "\n",
       "   Sp_gain/loss  Se_new_res  Se_allowed_err  Se_ref_result  Se_gain/loss  \\\n",
       "0         -0.01        0.66          0.0194         0.7206       -0.0606   \n",
       "1         -0.08        0.85          0.0129         0.9100       -0.0600   \n",
       "2         -0.01        0.75          0.0571         0.8400       -0.0900   \n",
       "3          0.01        0.88          0.0000         1.0000       -0.1200   \n",
       "4         -0.14        0.82          0.0194         0.7206        0.0994   \n",
       "5         -0.07        0.87          0.0129         0.9100       -0.0400   \n",
       "6         -0.15        0.85          0.0571         0.8400        0.0100   \n",
       "7          0.19        0.75          0.0000         1.0000       -0.2500   \n",
       "\n",
       "   Acc_new_res  Acc_allowed_err  Acc_ref_result  Acc_gain/loss  \n",
       "0         0.85            0.000            0.87          -0.02  \n",
       "1         0.77            0.010            0.85          -0.08  \n",
       "2         0.85            0.040            0.88          -0.03  \n",
       "3         0.66            0.057            0.68          -0.02  \n",
       "4         0.78            0.000            0.87          -0.09  \n",
       "5         0.78            0.010            0.85          -0.07  \n",
       "6         0.76            0.040            0.88          -0.12  \n",
       "7         0.78            0.057            0.68           0.10  "
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results = results.append(naive_gaussian_m3, ignore_index=True)\n",
    "results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3. LightGBM"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.1. Imbalanced Model towards nB compounds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[251  15]\n",
      " [ 15  53]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.94      0.94      0.94       266\n",
      "           1       0.78      0.78      0.78        68\n",
      "\n",
      "    accuracy                           0.91       334\n",
      "   macro avg       0.86      0.86      0.86       334\n",
      "weighted avg       0.91      0.91      0.91       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import lightgbm as lgb\n",
    "model = lgb.LGBMClassifier().fit(X_train, y_train)\n",
    "y_pred = model.predict(X_test) \n",
    "print(confusion_matrix(y_test,y_pred))\n",
    "print(classification_report(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this is the new result\n",
    "M1_sp_lightgbm=0.94\n",
    "M1_se_lightgbm=0.78\n",
    "M1_acc_lightgbm=0.91\n",
    "# this is the allowed error:\n",
    "M1_err_sp=0.00\n",
    "M1_err_se=0.0194\n",
    "M1_err_acc=0.00\n",
    "# this is the reference result:\n",
    "M1_sp=0.91\n",
    "M1_se=0.7206\n",
    "M1_acc=0.87\n",
    "# this is the gain or loss: new result minus reference result\n",
    "M1_sp_lightgbm_eval=(M1_sp_lightgbm-M1_sp)\n",
    "M1_se_lightgbm_eval=(M1_se_lightgbm-M1_se)\n",
    "M1_acc_lightgbm_eval=(M1_acc_lightgbm-M1_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "lightgbm_m1 = {'Model':\"M1:LightGBM\", 'Sp_new_res':M1_sp_lightgbm, 'Sp_allowed_err':M1_err_sp, 'Sp_ref_result':M1_sp, 'Sp_gain/loss':M1_sp_lightgbm_eval, 'Se_new_res':M1_se_lightgbm, 'Se_allowed_err':M1_err_se, 'Se_ref_result':M1_se, 'Se_gain/loss':M1_se_lightgbm_eval, 'Acc_new_res':M1_acc_lightgbm, 'Acc_allowed_err':M1_err_acc, 'Acc_ref_result':M1_acc, 'Acc_gain/loss':M1_acc_lightgbm_eval}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = results.append(lightgbm_m1, ignore_index=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.2. Balanced Model "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0    532\n",
       "1    141\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train1.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0    141\n",
       "1    141\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "positive_cases = train1[train1['BCF'] == 1]\n",
    "negative_sample = train1[train1['BCF'] == 0].sample(141, random_state=0)\n",
    "train1_balanced = pd.concat([positive_cases,negative_sample])\n",
    "train1_balanced.sort_index(inplace=True)\n",
    "\n",
    "X_train_bal = train1_balanced.drop(['BCF'], axis=1)\n",
    "y_train_bal = train1_balanced['BCF']\n",
    "\n",
    "# AFTER\n",
    "train1_balanced.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[222  44]\n",
      " [  4  64]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.98      0.83      0.90       266\n",
      "           1       0.59      0.94      0.73        68\n",
      "\n",
      "    accuracy                           0.86       334\n",
      "   macro avg       0.79      0.89      0.81       334\n",
      "weighted avg       0.90      0.86      0.87       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import lightgbm as lgb\n",
    "model = lgb.LGBMClassifier().fit(X_train_bal, y_train_bal)\n",
    "y_pred = model.predict(X_test) \n",
    "print(confusion_matrix(y_test,y_pred))\n",
    "print(classification_report(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this is the new result\n",
    "M2_sp_lightgbm=0.83\n",
    "M2_se_lightgbm=0.94\n",
    "M2_acc_lightgbm=0.86\n",
    "# this is the allowed error:\n",
    "M2_err_sp=0.00\n",
    "M2_err_se=0.0129\n",
    "M2_err_acc=0.01\n",
    "# this is the reference result:\n",
    "M2_sp=0.83\n",
    "M2_se=0.91\n",
    "M2_acc=0.85\n",
    "# this is the gain or loss: new result minus reference result\n",
    "M2_sp_lightgbm_eval=(M2_sp_lightgbm-M2_sp)\n",
    "M2_se_lightgbm_eval=(M2_se_lightgbm-M2_se)\n",
    "M2_acc_lightgbm_eval=(M2_acc_lightgbm-M2_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "lightgbm_m2 = {'Model':\"M2:LightGBM\", 'Sp_new_res':M2_sp_lightgbm, 'Sp_allowed_err':M2_err_sp, 'Sp_ref_result':M2_sp, 'Sp_gain/loss':M2_sp_lightgbm_eval, 'Se_new_res':M2_se_lightgbm, 'Se_allowed_err':M2_err_se, 'Se_ref_result':M2_se, 'Se_gain/loss':M2_se_lightgbm_eval, 'Acc_new_res':M2_acc_lightgbm, 'Acc_allowed_err':M2_err_acc, 'Acc_ref_result':M2_acc, 'Acc_gain/loss':M2_acc_lightgbm_eval}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = results.append(lightgbm_m2, ignore_index=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.2.2. Balanced Model - SMOTE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "import imblearn\n",
    "from imblearn.over_sampling import SMOTE\n",
    "smote = SMOTE(random_state=0)\n",
    "X_train_resampled, y_train_resampled = smote.fit_sample(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[246  20]\n",
      " [ 12  56]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.95      0.92      0.94       266\n",
      "           1       0.74      0.82      0.78        68\n",
      "\n",
      "    accuracy                           0.90       334\n",
      "   macro avg       0.85      0.87      0.86       334\n",
      "weighted avg       0.91      0.90      0.91       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "import lightgbm as lgb\n",
    "model = lgb.LGBMClassifier().fit(X_train_resampled, y_train_resampled)\n",
    "y_pred = model.predict(X_test) \n",
    "print(confusion_matrix(y_test,y_pred))\n",
    "print(classification_report(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this is the new result\n",
    "M2_2_sp_lightgbm=0.92\n",
    "M2_2_se_lightgbm=0.82\n",
    "M2_2_acc_lightgbm=0.90\n",
    "# this is the allowed error:\n",
    "M2_2_err_sp=0.06\n",
    "M2_2_err_se=0.0571\n",
    "M2_2_err_acc=0.04\n",
    "# this is the reference result:\n",
    "M2_2_sp=0.89\n",
    "M2_2_se=0.84\n",
    "M2_2_acc=0.88\n",
    "# this is the gain or loss: new result minus reference result\n",
    "M2_2_sp_lightgbm_eval=(M2_2_sp_lightgbm-M2_2_sp)\n",
    "M2_2_se_lightgbm_eval=(M2_2_se_lightgbm-M2_2_se)\n",
    "M2_2_acc_lightgbm_eval=(M2_2_acc_lightgbm-M2_2_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "lightgbm_m2_2 = {'Model':\"M2_2:LightGBM\", 'Sp_new_res':M2_2_sp_lightgbm, 'Sp_allowed_err':M2_2_err_sp, 'Sp_ref_result':M2_2_sp, 'Sp_gain/loss':M2_2_sp_lightgbm_eval, 'Se_new_res':M2_2_se_lightgbm, 'Se_allowed_err':M2_2_err_se, 'Se_ref_result':M2_2_se, 'Se_gain/loss':M2_2_se_lightgbm_eval, 'Acc_new_res':M2_2_acc_lightgbm, 'Acc_allowed_err':M2_2_err_acc, 'Acc_ref_result':M2_2_acc, 'Acc_gain/loss':M2_2_acc_lightgbm_eval}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = results.append(lightgbm_m2_2, ignore_index=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3.3. Imbalanced Model towards B compounds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0    532\n",
       "1    141\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 71,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train1.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "BCF\n",
       "0     28\n",
       "1    107\n",
       "Name: BCF, dtype: int64"
      ]
     },
     "execution_count": 72,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "positive_cases = train1[train1['BCF'] == 1].sample(107, random_state=0)\n",
    "negative_sample = train1[train1['BCF'] == 0].sample(28, random_state=0)\n",
    "train1_imbalanced = pd.concat([positive_cases,negative_sample])\n",
    "train1_imbalanced.sort_index(inplace=True)\n",
    "\n",
    "X_train_imbal = train1_imbalanced.drop(['BCF'], axis=1)\n",
    "y_train_imbal = train1_imbalanced['BCF']\n",
    "# AFTER\n",
    "train1_imbalanced.groupby(['BCF'])['BCF'].count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[161 105]\n",
      " [  0  68]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       1.00      0.61      0.75       266\n",
      "           1       0.39      1.00      0.56        68\n",
      "\n",
      "    accuracy                           0.69       334\n",
      "   macro avg       0.70      0.80      0.66       334\n",
      "weighted avg       0.88      0.69      0.72       334\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import classification_report, confusion_matrix\n",
    "import lightgbm as lgb\n",
    "model = lgb.LGBMClassifier().fit(X_train_imbal, y_train_imbal)\n",
    "y_pred = model.predict(X_test) \n",
    "print(confusion_matrix(y_test,y_pred))\n",
    "print(classification_report(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this is the new result\n",
    "M3_sp_lightgbm=0.61\n",
    "M3_se_lightgbm=1.00\n",
    "M3_acc_lightgbm=0.69\n",
    "# this is the allowed error:\n",
    "M3_err_sp=0.0692\n",
    "M3_err_se=0.00\n",
    "M3_err_acc=0.057\n",
    "# this is the reference result:\n",
    "M3_sp=0.6\n",
    "M3_se=1.00\n",
    "M3_acc=0.68\n",
    "# this is the gain or loss: new result minus reference result\n",
    "M3_sp_lightgbm_eval=(M3_sp_lightgbm-M3_sp)\n",
    "M3_se_lightgbm_eval=(M3_se_lightgbm-M3_se)\n",
    "M3_acc_lightgbm_eval=(M3_acc_lightgbm-M3_acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "lightgbm_m3 = {'Model':\"M3:LightGBM\", 'Sp_new_res':M3_sp_lightgbm, 'Sp_allowed_err':M3_err_sp, 'Sp_ref_result':M3_sp, 'Sp_gain/loss':M3_sp_lightgbm_eval, 'Se_new_res':M3_se_lightgbm, 'Se_allowed_err':M3_err_se, 'Se_ref_result':M3_se, 'Se_gain/loss':M3_se_lightgbm_eval, 'Acc_new_res':M3_acc_lightgbm, 'Acc_allowed_err':M3_err_acc, 'Acc_ref_result':M3_acc, 'Acc_gain/loss':M3_acc_lightgbm_eval}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Model</th>\n",
       "      <th>Sp_new_res</th>\n",
       "      <th>Sp_allowed_err</th>\n",
       "      <th>Sp_ref_result</th>\n",
       "      <th>Sp_gain/loss</th>\n",
       "      <th>Se_new_res</th>\n",
       "      <th>Se_allowed_err</th>\n",
       "      <th>Se_ref_result</th>\n",
       "      <th>Se_gain/loss</th>\n",
       "      <th>Acc_new_res</th>\n",
       "      <th>Acc_allowed_err</th>\n",
       "      <th>Acc_ref_result</th>\n",
       "      <th>Acc_gain/loss</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>M1:SVM_linear</td>\n",
       "      <td>0.90</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.91</td>\n",
       "      <td>-0.01</td>\n",
       "      <td>0.66</td>\n",
       "      <td>0.0194</td>\n",
       "      <td>0.7206</td>\n",
       "      <td>-0.0606</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.87</td>\n",
       "      <td>-0.02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>M2:SVM_linear</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.83</td>\n",
       "      <td>-0.08</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.0129</td>\n",
       "      <td>0.9100</td>\n",
       "      <td>-0.0600</td>\n",
       "      <td>0.77</td>\n",
       "      <td>0.010</td>\n",
       "      <td>0.85</td>\n",
       "      <td>-0.08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>M2_2:SVM_linear</td>\n",
       "      <td>0.88</td>\n",
       "      <td>0.0600</td>\n",
       "      <td>0.89</td>\n",
       "      <td>-0.01</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.0571</td>\n",
       "      <td>0.8400</td>\n",
       "      <td>-0.0900</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.040</td>\n",
       "      <td>0.88</td>\n",
       "      <td>-0.03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>M3:SVM_linear</td>\n",
       "      <td>0.61</td>\n",
       "      <td>0.0692</td>\n",
       "      <td>0.60</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.88</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>1.0000</td>\n",
       "      <td>-0.1200</td>\n",
       "      <td>0.66</td>\n",
       "      <td>0.057</td>\n",
       "      <td>0.68</td>\n",
       "      <td>-0.02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>M1:Naive_gaussian</td>\n",
       "      <td>0.77</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.91</td>\n",
       "      <td>-0.14</td>\n",
       "      <td>0.82</td>\n",
       "      <td>0.0194</td>\n",
       "      <td>0.7206</td>\n",
       "      <td>0.0994</td>\n",
       "      <td>0.78</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.87</td>\n",
       "      <td>-0.09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>M2:Naive_gaussian</td>\n",
       "      <td>0.76</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.83</td>\n",
       "      <td>-0.07</td>\n",
       "      <td>0.87</td>\n",
       "      <td>0.0129</td>\n",
       "      <td>0.9100</td>\n",
       "      <td>-0.0400</td>\n",
       "      <td>0.78</td>\n",
       "      <td>0.010</td>\n",
       "      <td>0.85</td>\n",
       "      <td>-0.07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>M2_2:Naive_gaussian</td>\n",
       "      <td>0.74</td>\n",
       "      <td>0.0600</td>\n",
       "      <td>0.89</td>\n",
       "      <td>-0.15</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.0571</td>\n",
       "      <td>0.8400</td>\n",
       "      <td>0.0100</td>\n",
       "      <td>0.76</td>\n",
       "      <td>0.040</td>\n",
       "      <td>0.88</td>\n",
       "      <td>-0.12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>M3:Naive_gaussian</td>\n",
       "      <td>0.79</td>\n",
       "      <td>0.0692</td>\n",
       "      <td>0.60</td>\n",
       "      <td>0.19</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>1.0000</td>\n",
       "      <td>-0.2500</td>\n",
       "      <td>0.78</td>\n",
       "      <td>0.057</td>\n",
       "      <td>0.68</td>\n",
       "      <td>0.10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>M1:LightGBM</td>\n",
       "      <td>0.94</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.91</td>\n",
       "      <td>0.03</td>\n",
       "      <td>0.78</td>\n",
       "      <td>0.0194</td>\n",
       "      <td>0.7206</td>\n",
       "      <td>0.0594</td>\n",
       "      <td>0.91</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.87</td>\n",
       "      <td>0.04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>M2:LightGBM</td>\n",
       "      <td>0.83</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.83</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.94</td>\n",
       "      <td>0.0129</td>\n",
       "      <td>0.9100</td>\n",
       "      <td>0.0300</td>\n",
       "      <td>0.86</td>\n",
       "      <td>0.010</td>\n",
       "      <td>0.85</td>\n",
       "      <td>0.01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>M2_2:LightGBM</td>\n",
       "      <td>0.92</td>\n",
       "      <td>0.0600</td>\n",
       "      <td>0.89</td>\n",
       "      <td>0.03</td>\n",
       "      <td>0.82</td>\n",
       "      <td>0.0571</td>\n",
       "      <td>0.8400</td>\n",
       "      <td>-0.0200</td>\n",
       "      <td>0.90</td>\n",
       "      <td>0.040</td>\n",
       "      <td>0.88</td>\n",
       "      <td>0.02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>M3:LightGBM</td>\n",
       "      <td>0.61</td>\n",
       "      <td>0.0692</td>\n",
       "      <td>0.60</td>\n",
       "      <td>0.01</td>\n",
       "      <td>1.00</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>1.0000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.69</td>\n",
       "      <td>0.057</td>\n",
       "      <td>0.68</td>\n",
       "      <td>0.01</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  Model  Sp_new_res  Sp_allowed_err  Sp_ref_result  \\\n",
       "0         M1:SVM_linear        0.90          0.0000           0.91   \n",
       "1         M2:SVM_linear        0.75          0.0000           0.83   \n",
       "2       M2_2:SVM_linear        0.88          0.0600           0.89   \n",
       "3         M3:SVM_linear        0.61          0.0692           0.60   \n",
       "4     M1:Naive_gaussian        0.77          0.0000           0.91   \n",
       "5     M2:Naive_gaussian        0.76          0.0000           0.83   \n",
       "6   M2_2:Naive_gaussian        0.74          0.0600           0.89   \n",
       "7     M3:Naive_gaussian        0.79          0.0692           0.60   \n",
       "8           M1:LightGBM        0.94          0.0000           0.91   \n",
       "9           M2:LightGBM        0.83          0.0000           0.83   \n",
       "10        M2_2:LightGBM        0.92          0.0600           0.89   \n",
       "11          M3:LightGBM        0.61          0.0692           0.60   \n",
       "\n",
       "    Sp_gain/loss  Se_new_res  Se_allowed_err  Se_ref_result  Se_gain/loss  \\\n",
       "0          -0.01        0.66          0.0194         0.7206       -0.0606   \n",
       "1          -0.08        0.85          0.0129         0.9100       -0.0600   \n",
       "2          -0.01        0.75          0.0571         0.8400       -0.0900   \n",
       "3           0.01        0.88          0.0000         1.0000       -0.1200   \n",
       "4          -0.14        0.82          0.0194         0.7206        0.0994   \n",
       "5          -0.07        0.87          0.0129         0.9100       -0.0400   \n",
       "6          -0.15        0.85          0.0571         0.8400        0.0100   \n",
       "7           0.19        0.75          0.0000         1.0000       -0.2500   \n",
       "8           0.03        0.78          0.0194         0.7206        0.0594   \n",
       "9           0.00        0.94          0.0129         0.9100        0.0300   \n",
       "10          0.03        0.82          0.0571         0.8400       -0.0200   \n",
       "11          0.01        1.00          0.0000         1.0000        0.0000   \n",
       "\n",
       "    Acc_new_res  Acc_allowed_err  Acc_ref_result  Acc_gain/loss  \n",
       "0          0.85            0.000            0.87          -0.02  \n",
       "1          0.77            0.010            0.85          -0.08  \n",
       "2          0.85            0.040            0.88          -0.03  \n",
       "3          0.66            0.057            0.68          -0.02  \n",
       "4          0.78            0.000            0.87          -0.09  \n",
       "5          0.78            0.010            0.85          -0.07  \n",
       "6          0.76            0.040            0.88          -0.12  \n",
       "7          0.78            0.057            0.68           0.10  \n",
       "8          0.91            0.000            0.87           0.04  \n",
       "9          0.86            0.010            0.85           0.01  \n",
       "10         0.90            0.040            0.88           0.02  \n",
       "11         0.69            0.057            0.68           0.01  "
      ]
     },
     "execution_count": 76,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results = results.append(lightgbm_m3, ignore_index=True)\n",
    "results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}