Task-1.lyx

#LyX 2.3 created this file. For more info see http://www.lyx.org/
\lyxformat 544
\begin_document
\begin_header
\save_transient_properties true
\origin unavailable
\textclass article
\use_default_options true
\begin_modules
knitr
\end_modules
\maintain_unincluded_children false
\language english
\language_package default
\inputencoding auto
\fontencoding global
\font_roman "default" "default"
\font_sans "default" "default"
\font_typewriter "default" "default"
\font_math "auto" "auto"
\font_default_family default
\use_non_tex_fonts false
\font_sc false
\font_osf false
\font_sf_scale 100 100
\font_tt_scale 100 100
\use_microtype false
\use_dash_ligatures true
\graphics default
\default_output_format default
\output_sync 0
\bibtex_command default
\index_command default
\paperfontsize default
\spacing single
\use_hyperref false
\papersize default
\use_geometry false
\use_package amsmath 1
\use_package amssymb 1
\use_package cancel 1
\use_package esint 1
\use_package mathdots 1
\use_package mathtools 1
\use_package mhchem 1
\use_package stackrel 1
\use_package stmaryrd 1
\use_package undertilde 1
\cite_engine basic
\cite_engine_type default
\biblio_style plain
\use_bibtopic false
\use_indices false
\paperorientation portrait
\suppress_date false
\justification true
\use_refstyle 1
\use_minted 0
\index Index
\shortcut idx
\color #008000
\end_index
\secnumdepth 3
\tocdepth 3
\paragraph_separation indent
\paragraph_indentation default
\is_math_indent 0
\math_numbering_side default
\quotes_style english
\dynamic_quotes 0
\papercolumns 1
\papersides 1
\paperpagestyle default
\tracking_changes false
\output_changes false
\html_math_output 0
\html_css_as_file 0
\html_be_strict false
\end_header

\begin_body

\begin_layout Title
LGM VIP Data Science Internship
\begin_inset Newline newline
\end_inset

Beginner Task-1: Iris Flowers Classification ML Project
\end_layout

\begin_layout Author
Name: Debartha Paul
\end_layout

\begin_layout Section*
Description:
\end_layout

\begin_layout Standard
The iris flowers dataset contains numeric attributes.
 We have the attributes as:
\end_layout

\begin_layout Enumerate
sepal length in cm 
\end_layout

\begin_layout Enumerate
sepal width in cm 
\end_layout

\begin_layout Enumerate
petal length in cm 
\end_layout

\begin_layout Enumerate
petal width in cm 
\end_layout

\begin_layout Enumerate
class: 
\end_layout

\begin_deeper
\begin_layout Enumerate
-- Iris Setosa 
\end_layout

\begin_layout Enumerate
-- Iris Versicolour 
\end_layout

\begin_layout Enumerate
-- Iris Virginica
\end_layout

\end_deeper
\begin_layout Standard
We have to classify them into different categories using ML.
\end_layout

\begin_layout Section*
Importing libraries and visualising the data
\end_layout

\begin_layout Standard
We first load the libraries required for our work and then we read the dataset
\end_layout

\begin_layout Standard
\begin_inset ERT
status open

\begin_layout Plain Layout

\end_layout

\begin_layout Plain Layout

<<>>=
\end_layout

\begin_layout Plain Layout

library(class)#for the KNN function
\end_layout

\begin_layout Plain Layout

set.seed(200)#setting the seed
\end_layout

\begin_layout Plain Layout

df<-read.table('D:
\backslash

\backslash
Important Documents
\backslash

\backslash
Internship
\backslash

\backslash
LGM-VIP Aug21
\backslash

\backslash
Task-1
\backslash

\backslash
iris.data',
\end_layout

\begin_layout Plain Layout

				sep=',',header=F)#reading the file
\end_layout

\begin_layout Plain Layout

head(df)#a sneak peek into the dataset
\end_layout

\begin_layout Plain Layout

names(df)<-c('sepal.len','sepal.wd','petal.len',
\end_layout

\begin_layout Plain Layout

			'petal.wd','class')#setting the names of the columns
\end_layout

\begin_layout Plain Layout

dim(df)#the dimensions of the dataset
\end_layout

\begin_layout Plain Layout

summary(df)#a brief summary of the dataset
\end_layout

\begin_layout Plain Layout

classno<-ifelse(df$class=='Iris-setosa',1,
\end_layout

\begin_layout Plain Layout

				ifelse(df$class=='Iris-virginica',2,3))#factoring on the basis of class
\end_layout

\begin_layout Plain Layout

\end_layout

\begin_layout Plain Layout

pairs(df[,-5],col=c('#943737','#599437','#375f94')[classno],
\end_layout

\begin_layout Plain Layout

		main='Pairs plot of Iris',pch=16,oma=c(3,5,5,12))#plots of the data by
 various attributes
\end_layout

\begin_layout Plain Layout

par(xpd=T)
\end_layout

\begin_layout Plain Layout

legend("bottomright",fill=c('#943737','#599437','#375f94'),
\end_layout

\begin_layout Plain Layout

		legend=c('Iris-setosa','Iris-virginica','Iris-versicolor'),
\end_layout

\begin_layout Plain Layout

		cex=0.5,title='Class')
\end_layout

\begin_layout Plain Layout

@
\end_layout

\begin_layout Plain Layout

\end_layout

\end_inset


\end_layout

\begin_layout Section*
Splitting the dataset into training and test set
\end_layout

\begin_layout Standard
We next split our dataset into training and testing sets.
 
\end_layout

\begin_layout Standard
\begin_inset ERT
status open

\begin_layout Plain Layout

\end_layout

\begin_layout Plain Layout

<<>>=
\end_layout

\begin_layout Plain Layout

t_sample<-sample(nrow(df),0.8*nrow(df))	#contains the row numbers of the
 training set
\end_layout

\begin_layout Plain Layout

train<-df[t_sample,];test<-df[-t_sample,]
\end_layout

\begin_layout Plain Layout

@
\end_layout

\begin_layout Plain Layout

\end_layout

\end_inset


\end_layout

\begin_layout Standard
We will use the K nearest neighbours to classify the dataset.
\end_layout

\begin_layout Section*
Value of k
\end_layout

\begin_layout Standard
First, we obtain the the number of neighbours that are to be considered.
 We'll choose the 
\family typewriter
k
\family default
 which has the max accuracy.
 We start by taking 
\family typewriter
k
\family default
 as 1 and then simulate the accuracy incrementing 
\family typewriter
k
\family default
 in each step.
\end_layout

\begin_layout Standard
\begin_inset ERT
status open

\begin_layout Plain Layout

\end_layout

\begin_layout Plain Layout

<<>>=
\end_layout

\begin_layout Plain Layout

model_knn<-list();accuracy_knn<-numeric()
\end_layout

\begin_layout Plain Layout

for(i in 1:20)
\end_layout

\begin_layout Plain Layout

{
\end_layout

\begin_layout Plain Layout

	model_knn[[i]]<-knn(train[,-5],test[,-5],train$class,k=i,prob=T)
\end_layout

\begin_layout Plain Layout

	accuracy_knn[i]<-sum(model_knn[[i]]==test$class)/length(test$class)
\end_layout

\begin_layout Plain Layout

}
\end_layout

\begin_layout Plain Layout

plot(1:20,accuracy_knn,type='b',main='Accuracy vs.
 k',ylab='Accuracy',xlab='k',col='#33435c',pch=16)
\end_layout

\begin_layout Plain Layout

@
\end_layout

\begin_layout Plain Layout

\end_layout

\end_inset


\end_layout

\begin_layout Standard
From the plot, we find that there are many values of 
\family typewriter
k
\family default
 which have maximum accuracy among all the values of 
\family typewriter
k
\family default
.
\end_layout

\begin_layout Standard
We proceed to find the maximum accuracy.
\end_layout

\begin_layout Standard
\begin_inset ERT
status open

\begin_layout Plain Layout

\end_layout

\begin_layout Plain Layout

<<>>=
\end_layout

\begin_layout Plain Layout

max(accuracy_knn)
\end_layout

\begin_layout Plain Layout

@
\end_layout

\begin_layout Plain Layout

\end_layout

\end_inset


\end_layout

\begin_layout Standard
Thus, we find that the max accuracy that was obtained by this method is
 
\begin_inset Formula $96.67\%$
\end_inset

.
\end_layout

\begin_layout Standard
Further, we find the least vaue of 
\family typewriter
k
\family default
 with maximum accuracy.
 This is due to the fact that a greater value of 
\family typewriter
k
\family default
 may lead to overfitting of the data.
\end_layout

\begin_layout Standard
\begin_inset ERT
status open

\begin_layout Plain Layout

\end_layout

\begin_layout Plain Layout

<<>>=
\end_layout

\begin_layout Plain Layout

k<-min(which(accuracy_knn==max(accuracy_knn)))#min value of k which has
 the max accuracy
\end_layout

\begin_layout Plain Layout

k
\end_layout

\begin_layout Plain Layout

@
\end_layout

\begin_layout Plain Layout

\end_layout

\end_inset


\end_layout

\begin_layout Standard
Thus, we find the value of 
\family typewriter
k
\family default
 as 
\begin_inset Formula $10$
\end_inset

 with an accuracy level of 
\begin_inset Formula $96.67\%$
\end_inset

.
\end_layout

\begin_layout Section*
Predictions
\end_layout

\begin_layout Standard
We finally classify the iris dataset into the respective classes.
\end_layout

\begin_layout Standard
\begin_inset ERT
status open

\begin_layout Plain Layout

\end_layout

\begin_layout Plain Layout

<<>>=
\end_layout

\begin_layout Plain Layout

final_model <- knn(train[,-5], test[,-5], train$class, k, prob=TRUE)
\end_layout

\begin_layout Plain Layout

model_tab <- table(test$class, final_model)
\end_layout

\begin_layout Plain Layout

model_tab
\end_layout

\begin_layout Plain Layout

@
\end_layout

\begin_layout Plain Layout

\end_layout

\end_inset


\end_layout

\begin_layout Standard
We find that, while 
\family typewriter
Iris-setosa
\family default
 was correctly classified, 
\family typewriter
Iris-versicolor
\family default
 and 
\family typewriter
Iris-virginica
\family default
 were not classified perfectly, as one of them was interchanged.
\end_layout

\begin_layout Standard
The below dataframe gives the details of the test set along with the original
 and predicted classes:
\end_layout

\begin_layout Standard
\begin_inset ERT
status open

\begin_layout Plain Layout

\end_layout

\begin_layout Plain Layout

<<>>=
\end_layout

\begin_layout Plain Layout

cbind(test,final_model)
\end_layout

\begin_layout Plain Layout

@
\end_layout

\begin_layout Plain Layout

\end_layout

\end_inset


\end_layout

\end_body
\end_document