Predicting Vehicle Insurance Interest Using Logistic Regression and Decision Trees

Abstract
A insurance company who want to know which customer will but their vehicle insurance in
future and who will not conduct a data analysis task on a huge dataset of more than 380 thousand
customers. Logistic regression is used by them to analyze the data set. This report discuss about
the logistic regression algorithm, its working methodology, type of logistic regression and its
counterpart decision tree. Logistic regression is good algorithm in terms of its easy to
implementation and simplicity but require a linear relationship between the attributes of data set
which is almost not possible in present world. Decision tree algorithm can be counterpart of this
algorithm because of it’s not tendency on data linearity. Decision tree is also best because tis
output can be visualize with the logic of splitting.

Table of Contents
Abstract............................................................................................................................................2
Introduction......................................................................................................................................5
Data set............................................................................................................................................5
Data set variables.........................................................................................................................6
Data Mining.....................................................................................................................................8
Logistic regression.......................................................................................................................8
Logistic regression property....................................................................................................9
Sigmoid Function.....................................................................................................................9
Types of Logistic Regression................................................................................................10
Linear regression vs logistic regression.................................................................................11
MLE vs OLS..........................................................................................................................11
Advantages of logistic regression..........................................................................................12
Dis-advantages of logistic regression....................................................................................12
Decision Tree Algorithm...........................................................................................................12
Decision Tree algorithm working..........................................................................................13
Recursive Binary Splitting.....................................................................................................14
Conclusion.....................................................................................................................................17
Reference.......................................................................................................................................18

List of Figures
Figure 1 - Sigmoid Curve..............................................................................................................10
Figure 2 - Linear and Logistic Regression Curves........................................................................11
Figure 3 - Flowchart type of Decision Tree...................................................................................13
Figure 4 - Decision Tree working Methodology...........................................................................14
List of Table
Table 1 - Description of data set variables......................................................................................6