Please use this identifier to cite or link to this item:
http://hdl.handle.net/10603/444251
Title: | Machine learning model for customer attrition prediction in motor insurance sector |
Researcher: | DEEPTHI DAS |
Guide(s): | RAJU RAMAKRISHNA GONDKAR |
Keywords: | Computer Science Computer Science Interdisciplinary Applications Engineering and Technology |
University: | CMR University |
Completed Date: | 2021 |
Abstract: | One of the major challenging problem in the different industrial sector is to forecast the attrition of customers. It is observed that the vehicle insurance sector faces this as a major issues as there are many insurance companies and therefore a lot of competition exists in the market. There are lot of upgradations in the policies and rules which makes the process of retaining the customers as it plays a very vital role in the growth of the insurance companies. The aim of this research work is to find the class of customers who will churn and the class of customers who will not churn from the company by analyzing the various behaviors of the customers. In this work we also find the most significant attributes which helps in identifying the churners. newlineIn this research we have used a data set from a motor insurance company and the data set consists of 20,000 rows with 37 columns. The absent values in the data set are examined using Expectation Maximization algorithm. The available data is grouped according to the renewal of policies. To build the model the various attributes of the customers in the data set is analyzed and the best features are selected. Gaussian Mixture Model is used to group the customers according to their behaviors on the policies which are analyzed. The dependency rate of each variable is calculated and analyzed. A hybrid GWO-KELM algorithm is performed on the data to identify the customers who churn and the customers who will not churn. GWO algorithm helps in determining, exploring and analyzing the optimal feature. The results of this research study have shown an efficiency of the hybrid algorithm as prediction accuracy of 95%, precision of 97%, a recall of 91% and F-score of 94%. newline |
Pagination: | |
URI: | http://hdl.handle.net/10603/444251 |
Appears in Departments: | School of Engineering and Technology |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
80_recommendation.pdf | Attached File | 79.24 kB | Adobe PDF | View/Open |
chapter 1.pdf | 719.84 kB | Adobe PDF | View/Open | |
chapter 2.pdf | 236.88 kB | Adobe PDF | View/Open | |
chapter 3.pdf | 102.55 kB | Adobe PDF | View/Open | |
chapter 4.pdf | 230.72 kB | Adobe PDF | View/Open | |
chapter 5.pdf | 459.28 kB | Adobe PDF | View/Open | |
chapter 6.pdf | 95.85 kB | Adobe PDF | View/Open | |
declaration + list of content.pdf | 108.7 kB | Adobe PDF | View/Open | |
references.pdf | 210.16 kB | Adobe PDF | View/Open | |
title page.pdf | 74.29 kB | Adobe PDF | View/Open |
Items in Shodhganga are licensed under Creative Commons Licence Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).
Altmetric Badge: