The Auto Insurance Churn Using Machine Learning Models

A Machine Learning Approach to Predicting Auto Insurance Churn

Authors

  • Usman Ishtiaq Superior Univeristy

DOI:

https://doi.org/10.56536/jicet.v4i2.159

Keywords:

Machine Learning, Auto Insurance, Random Forest, LightGBM, XGBooster

Abstract

This paper dives into the existing churnScience prediction methods over different segments such as Web administrations, gaming, protections, and administration. For companies pointing at maintained development and expanded client maintenance, a vigorous client churn forecast is basic. Notwithstanding a company's benefit quality, recognizing focuses where clients suspend employing a benefit, item, or platform holds noteworthy significance. Typically, it is especially pivotal for different businesses, recognizing that holding current clients is regularly more cost-effective than acquiring unused ones. The essential commitment of this inquiry lies within the development of a proactive prescient demonstration able to predict clients who are likely to terminate employing a benefit, item, or stage. They consider utilizing machine learning strategies, particularly on auto protection churn information. Another outstanding commitment is the utilization of the XGBooster algorithm, Gradient Boosting, and Random Forest Tree for data labeling, pointing to improve accuracy. The data is apportioned into 70% for training and 30% for testing, with hyperparameter tuning utilized to optimize the machine learning demonstration. So, we can compare all the algorithms for better accuracy of data and obtain an improved accuracy is 88.95% and the F1 score is 94% respectively.

Downloads

Published

2024-10-15