forked from yuxiaowww/BDCI-2018-Supply-Chain-Demand-Forecast
-
Notifications
You must be signed in to change notification settings - Fork 0
/
merge-1111-提交.py
228 lines (205 loc) · 12.4 KB
/
merge-1111-提交.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
# -*- coding: utf-8 -*-
"""
Created on Tue Sep 18 16:26:28 2018
@author: yuwei
"""
import pandas as pd
'---------------------------------------------------1-----------------------------------------------------'
'第一步融合：规则+张浩lr更新+张浩lgb'
data1 = pd.read_csv('yw-rule.csv')
data2 = pd.read_csv('zh_lr_v1.csv')
data3 = pd.read_csv('zh_lgb_v1.csv')
data4 = pd.read_csv('zh_lgb_v3.csv')
#对data2归总和
w1 = 296264;w2 = 399559;w3 = 554991;w4 = 641003;w5 = 244596
w1 = w1/sum(data2['week1']);w2 = w2/sum(data2['week2']);w3 = w3/sum(data2['week3']);w4 = w4/sum(data2['week4']);w5 = w5/sum(data2['week5'])
data2['week1'] = data2['week1'].map(lambda x: w1*x)
data2['week2'] = data2['week2'].map(lambda x: w2*x)
data2['week3'] = data2['week3'].map(lambda x: w3*x)
data2['week4'] = data2['week4'].map(lambda x: w4*x)
data2['week5'] = data2['week5'].map(lambda x: w5*x)
#对data3归总和
w1 = 296264;w2 = 399559;w3 = 554991;w4 = 641003;w5 = 244596
w1 = w1/sum(data3['week1']);w2 = w2/sum(data3['week2']);w3 = w3/sum(data3['week3']);w4 = w4/sum(data3['week4']);w5 = w5/sum(data3['week5'])
data3['week1'] = data3['week1'].map(lambda x: w1*x)
data3['week2'] = data3['week2'].map(lambda x: w2*x)
data3['week3'] = data3['week3'].map(lambda x: w3*x)
data3['week4'] = data3['week4'].map(lambda x: w4*x)
data3['week5'] = data3['week5'].map(lambda x: w5*x)
#对data4归总和
w1 = 296264;w2 = 399559;w3 = 554991;w4 = 641003;w5 = 244596
w1 = w1/sum(data4['week1']);w2 = w2/sum(data4['week2']);w3 = w3/sum(data4['week3']);w4 = w4/sum(data4['week4']);w5 = w5/sum(data4['week5'])
data4['week1'] = data4['week1'].map(lambda x: w1*x)
data4['week2'] = data4['week2'].map(lambda x: w2*x)
data4['week3'] = data4['week3'].map(lambda x: w3*x)
data4['week4'] = data4['week4'].map(lambda x: w4*x)
data4['week5'] = data4['week5'].map(lambda x: w5*x)
data1.columns = ['sku_id','week1_1','week2_1','week3_1','week4_1','week5_1']
data2.columns = ['sku_id','week1_2','week2_2','week3_2','week4_2','week5_2']
data3.columns = ['sku_id','week1_3','week2_3','week3_3','week4_3','week5_3']
data4.columns = ['sku_id','week1_4','week2_4','week3_4','week4_4','week5_4']
data = pd.merge(data1,data2,on='sku_id',how='left')
data = pd.merge(data,data3,on='sku_id',how='left')
data = pd.merge(data,data4,on='sku_id',how='left')
data['week1'] = list(map(lambda x,y,z,t: x*0.48+y*0.32+z*0.1+t*0.1 ,data.week1_1,data.week1_2,data.week1_3,data.week1_4))
data['week2'] = list(map(lambda x,y,z,t: x*0.48+y*0.32+z*0.1+t*0.1 ,data.week2_1,data.week2_2,data.week2_3,data.week2_4))
data['week3'] = list(map(lambda x,y,z,t: x*0.48+y*0.32+z*0.1+t*0.1 ,data.week3_1,data.week3_2,data.week3_3,data.week3_4))
data['week4'] = list(map(lambda x,y,z,t: x*0.48+y*0.32+z*0.1+t*0.1 ,data.week4_1,data.week4_2,data.week4_3,data.week4_4))
data['week5'] = list(map(lambda x,y,z,t: x*0.48+y*0.32+z*0.1+t*0.1 ,data.week5_1,data.week5_2,data.week5_3,data.week5_4))
data = data[['sku_id','week1','week2','week3','week4','week5']]
data['week1'] = data['week1'].map(lambda x: 1*x)
data['week2'] = data['week2'].map(lambda x: 1*x)
data['week3'] = data['week3'].map(lambda x: 1*x)
data['week4'] = data['week4'].map(lambda x: 1*x)
data['week5'] = data['week5'].map(lambda x: 1*x)
data.to_csv('merge1.csv',index=False)
'融合：张浩lgb+陈禹更新'
data1 = pd.read_csv('zh_xgb_v2.csv')
data2 = pd.read_csv('cy-result.csv')
data3 = pd.read_csv('zh_xgb_v1.csv')
data4 = pd.read_csv('zh_lgb_v4.csv')
#对data1归总和
w1 = 296264;w2 = 399559;w3 = 554991;w4 = 641003;w5 = 244596
w1 = w1/sum(data1['week1']);w2 = w2/sum(data1['week2']);w3 = w3/sum(data1['week3']);w4 = w4/sum(data1['week4']);w5 = w5/sum(data1['week5'])
data1['week1'] = data1['week1'].map(lambda x: w1*x)
data1['week2'] = data1['week2'].map(lambda x: w2*x)
data1['week3'] = data1['week3'].map(lambda x: w3*x)
data1['week4'] = data1['week4'].map(lambda x: w4*x)
data1['week5'] = data1['week5'].map(lambda x: w5*x)
#对data2归总和
w1 = 296264;w2 = 399559;w3 = 554991;w4 = 641003;w5 = 244596
w1 = w1/sum(data2['week1']);w2 = w2/sum(data2['week2']);w3 = w3/sum(data2['week3']);w4 = w4/sum(data2['week4']);w5 = w5/sum(data2['week5'])
data2['week1'] = data2['week1'].map(lambda x: w1*x)
data2['week2'] = data2['week2'].map(lambda x: w2*x)
data2['week3'] = data2['week3'].map(lambda x: w3*x)
data2['week4'] = data2['week4'].map(lambda x: w4*x)
data2['week5'] = data2['week5'].map(lambda x: w5*x)
#对data3归总和
w1 = 296264;w2 = 399559;w3 = 554991;w4 = 641003;w5 = 244596
w1 = w1/sum(data3['week1']);w2 = w2/sum(data3['week2']);w3 = w3/sum(data3['week3']);w4 = w4/sum(data3['week4']);w5 = w5/sum(data3['week5'])
data3['week1'] = data3['week1'].map(lambda x: w1*x)
data3['week2'] = data3['week2'].map(lambda x: w2*x)
data3['week3'] = data3['week3'].map(lambda x: w3*x)
data3['week4'] = data3['week4'].map(lambda x: w4*x)
data3['week5'] = data3['week5'].map(lambda x: w5*x)
#对data4归总和
w1 = 296264;w2 = 399559;w3 = 554991;w4 = 641003;w5 = 244596
w1 = w1/sum(data4['week1']);w2 = w2/sum(data4['week2']);w3 = w3/sum(data4['week3']);w4 = w4/sum(data4['week4']);w5 = w5/sum(data4['week5'])
data4['week1'] = data4['week1'].map(lambda x: w1*x)
data4['week2'] = data4['week2'].map(lambda x: w2*x)
data4['week3'] = data4['week3'].map(lambda x: w3*x)
data4['week4'] = data4['week4'].map(lambda x: w4*x)
data4['week5'] = data4['week5'].map(lambda x: w5*x)
data1.columns = ['sku_id','week1_1','week2_1','week3_1','week4_1','week5_1']
data2.columns = ['sku_id','week1_2','week2_2','week3_2','week4_2','week5_2']
data3.columns = ['sku_id','week1_3','week2_3','week3_3','week4_3','week5_3']
data4.columns = ['sku_id','week1_4','week2_4','week3_4','week4_4','week5_4']
data = pd.merge(data1,data2,on='sku_id',how='left')
data = pd.merge(data,data3,on='sku_id',how='left')
data = pd.merge(data,data4,on='sku_id',how='left')
data['week1'] = list(map(lambda x,y,z,t: x*0.65+y*0.19+z*0.09+t*0.07 ,data.week1_1,data.week1_2,data.week1_3,data.week1_4))
data['week2'] = list(map(lambda x,y,z,t: x*0.65+y*0.19+z*0.09+t*0.07 ,data.week2_1,data.week2_2,data.week2_3,data.week2_4))
data['week3'] = list(map(lambda x,y,z,t: x*0.65+y*0.19+z*0.09+t*0.07 ,data.week3_1,data.week3_2,data.week3_3,data.week3_4))
data['week4'] = list(map(lambda x,y,z,t: x*0.65+y*0.19+z*0.09+t*0.07 ,data.week4_1,data.week4_2,data.week4_3,data.week4_4))
data['week5'] = list(map(lambda x,y,z,t: x*0.65+y*0.19+z*0.09+t*0.07 ,data.week5_1,data.week5_2,data.week5_3,data.week5_4))
data = data[['sku_id','week1','week2','week3','week4','week5']]
data['week1'] = data['week1'].map(lambda x: 1*x)
data['week2'] = data['week2'].map(lambda x: 1*x)
data['week3'] = data['week3'].map(lambda x: 1*x)
data['week4'] = data['week4'].map(lambda x: 1*x)
data['week5'] = data['week5'].map(lambda x: 1*x)
data.to_csv('merge2.csv',index=False)
'融合：两部分融合结果再融合'
data1 = pd.read_csv('merge1.csv')
data2 = pd.read_csv('merge2.csv')
data1.columns = ['sku_id','week1_1','week2_1','week3_1','week4_1','week5_1']
data2.columns = ['sku_id','week1_2','week2_2','week3_2','week4_2','week5_2']
data = pd.merge(data1,data2,on='sku_id',how='left')
data['week1'] = list(map(lambda x,y: x*0.67+y*0.33 ,data.week1_1,data.week1_2))
data['week2'] = list(map(lambda x,y: x*0.67+y*0.33 ,data.week2_1,data.week2_2))
data['week3'] = list(map(lambda x,y: x*0.67+y*0.33 ,data.week3_1,data.week3_2))
data['week4'] = list(map(lambda x,y: x*0.67+y*0.33 ,data.week4_1,data.week4_2))
data['week5'] = list(map(lambda x,y: x*0.67+y*0.33 ,data.week5_1,data.week5_2))
data = data[['sku_id','week1','week2','week3','week4','week5']]
data['week1'] = data['week1'].map(lambda x: 1*x)
data['week2'] = data['week2'].map(lambda x: 1*x)
data['week3'] = data['week3'].map(lambda x: 1*x)
data['week4'] = data['week4'].map(lambda x: 1*x)
data['week5'] = data['week5'].map(lambda x: 1*x)
data.to_csv('merge3.csv',index=False)
------------------------------------------------------------------------------1
'第二步融合：三个lgb模型的融合+其中lgb更新+xwlr'
data1 = pd.read_csv('zh_lgb_v1.csv')
data2 = pd.read_csv('zh_lgb_v2.csv')
data3 = pd.read_csv('yw-lgb.csv')
data4 = pd.read_csv('zh_lr_v2.csv')
#对data1归总和
data1['week1'] = data1['week1'].map(lambda x: 1.15*x)
data1['week2'] = data1['week2'].map(lambda x: 1.30*x)
data1['week3'] = data1['week3'].map(lambda x: 1.75*x)
data1['week4'] = data1['week4'].map(lambda x: 1.30*x)
data1['week5'] = data1['week5'].map(lambda x: 0.85*x)
#对data2归总和
w1 = 298763;w2 = 396929;w3 = 553596;w4 = 630889;w5 = 242841
w1 = w1/sum(data2['week1']);w2 = w2/sum(data2['week2']);w3 = w3/sum(data2['week3']);w4 = w4/sum(data2['week4']);w5 = w5/sum(data2['week5'])
data2['week1'] = data2['week1'].map(lambda x: w1*x)
data2['week2'] = data2['week2'].map(lambda x: w2*x)
data2['week3'] = data2['week3'].map(lambda x: w3*x)
data2['week4'] = data2['week4'].map(lambda x: w4*x)
data2['week5'] = data2['week5'].map(lambda x: w5*x)
#对data3归总和
w1 = 298763;w2 = 396929;w3 = 553596;w4 = 630889;w5 = 242841
w1 = w1/sum(data3['week1']);w2 = w2/sum(data3['week2']);w3 = w3/sum(data3['week3']);w4 = w4/sum(data3['week4']);w5 = w5/sum(data3['week5'])
data3['week1'] = data3['week1'].map(lambda x: w1*x)
data3['week2'] = data3['week2'].map(lambda x: w2*x)
data3['week3'] = data3['week3'].map(lambda x: w3*x)
data3['week4'] = data3['week4'].map(lambda x: w4*x)
data3['week5'] = data3['week5'].map(lambda x: w5*x)
#对data4归总和
w1 = 298763;w2 = 396929;w3 = 553596;w4 = 630889;w5 = 242841
w1 = w1/sum(data4['week1']);w2 = w2/sum(data4['week2']);w3 = w3/sum(data4['week3']);w4 = w4/sum(data4['week4']);w5 = w5/sum(data4['week5'])
data4['week1'] = data4['week1'].map(lambda x: w1*x)
data4['week2'] = data4['week2'].map(lambda x: w2*x)
data4['week3'] = data4['week3'].map(lambda x: w3*x)
data4['week4'] = data4['week4'].map(lambda x: w4*x)
data4['week5'] = data4['week5'].map(lambda x: w5*x)
data1.columns = ['sku_id','week1_1','week2_1','week3_1','week4_1','week5_1']
data2.columns = ['sku_id','week1_2','week2_2','week3_2','week4_2','week5_2']
data3.columns = ['sku_id','week1_3','week2_3','week3_3','week4_3','week5_3']
data4.columns = ['sku_id','week1_4','week2_4','week3_4','week4_4','week5_4']
data = pd.merge(data1,data2,on='sku_id',how='left')
data = pd.merge(data,data3,on='sku_id',how='left')
data = pd.merge(data,data4,on='sku_id',how='left')
data['week1'] = list(map(lambda x,y,z,t: x*0.11+y*0.01+z*0.01+t*0.87 ,data.week1_1,data.week1_2,data.week1_3,data.week1_4))
data['week2'] = list(map(lambda x,y,z,t: x*0.11+y*0.01+z*0.01+t*0.87 ,data.week2_1,data.week2_2,data.week2_3,data.week2_4))
data['week3'] = list(map(lambda x,y,z,t: x*0.11+y*0.01+z*0.01+t*0.87 ,data.week3_1,data.week3_2,data.week3_3,data.week3_4))
data['week4'] = list(map(lambda x,y,z,t: x*0.11+y*0.01+z*0.01+t*0.87 ,data.week4_1,data.week4_2,data.week4_3,data.week4_4))
data['week5'] = list(map(lambda x,y,z,t: x*0.11+y*0.01+z*0.01+t*0.87 ,data.week5_1,data.week5_2,data.week5_3,data.week5_4))
data = data[['sku_id','week1','week2','week3','week4','week5']]
w1 = 297491;w2 = 398007;w3 = 553967;w4 = 635277;w5 = 243644
w1 = w1/sum(data['week1']);w2 = w2/sum(data['week2']);w3 = w3/sum(data['week3']);w4 = w4/sum(data['week4']);w5 = w5/sum(data['week5'])
data['week1'] = data['week1'].map(lambda x: w1*x)
data['week2'] = data['week2'].map(lambda x: w2*x)
data['week3'] = data['week3'].map(lambda x: w3*x)
data['week4'] = data['week4'].map(lambda x: w4*x)
data['week5'] = data['week5'].map(lambda x: w5*x)
data.to_csv('merge4.csv',index=False)
------------------------------------------------------------------------------2
'最后一步融合：两部分融合结果再融合'
data1 = pd.read_csv('merge3.csv')
data2 = pd.read_csv('merge4.csv')
data1.columns = ['sku_id','week1_1','week2_1','week3_1','week4_1','week5_1']
data2.columns = ['sku_id','week1_2','week2_2','week3_2','week4_2','week5_2']
data = pd.merge(data1,data2,on='sku_id',how='left')
data['week1'] = list(map(lambda x,y: x*0.18+y*0.82 ,data.week1_1,data.week1_2))
data['week2'] = list(map(lambda x,y: x*0.18+y*0.82 ,data.week2_1,data.week2_2))
data['week3'] = list(map(lambda x,y: x*0.18+y*0.82 ,data.week3_1,data.week3_2))
data['week4'] = list(map(lambda x,y: x*0.18+y*0.82 ,data.week4_1,data.week4_2))
data['week5'] = list(map(lambda x,y: x*0.18+y*0.82 ,data.week5_1,data.week5_2))
data = data[['sku_id','week1','week2','week3','week4','week5']]
data['week1'] = data['week1'].map(lambda x: 1.01*x)
data['week2'] = data['week2'].map(lambda x: 1.03*x)
data['week3'] = data['week3'].map(lambda x: 1.05*x)
data['week4'] = data['week4'].map(lambda x: 1.12*x)
data['week5'] = data['week5'].map(lambda x: 1.04*x)
data.to_csv('ans_merge_1111_B_lucky_复现.csv',index=False)