Development of customer response prediction model for personal loan using logistic regression
In general, financial service companies use mass media marketing strategies to offer and promote a product or service to their customers. This strategy is seen as less effective because differences in customer interests are not taken into account. With a large customer database, financial service co...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English |
| Published: |
2023
|
| Subjects: | |
| Online Access: | https://etd.uum.edu.my/10906/2/depositpermission_817737.pdf https://etd.uum.edu.my/10906/3/s817737_01.pdf https://etd.uum.edu.my/10906/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In general, financial service companies use mass media marketing strategies to offer and promote a product or service to their customers. This strategy is seen as less effective because differences in customer interests are not taken into account. With a large customer database, financial service companies have shifted from unproductive mass media marketing strategies to targeted marketing strategies by identifying customers who are likely to respond to more specific products and services. However, this strategy appears challenging when implemented manually through a large customer database. Therefore,
data mining techniques are used to develop accurate customer response forecasting
models. The goal of this study is to develop a customer response model to predict whether
a customer will respond to a personal loan product using logistic regression and decision
tree techniques. For modelling purposes, 10,952 customer data were collected from one
commercial bank in Malaysia. Factor analysis was conducted to identify inputs for the
model. Finally, logistic regression and decision tree techniques were used to develop the
customer response model. Different models from these two techniques were compared.
The best model for predicting customers who are likely to respond to personal loan
products was selected based on the highest accuracy, precision, and recall values. Both
models had the same accuracy of around 98%. However, decision tree performed slightly
better than regression, especially for 10-fold cross-validation. For 5-fold cross-validation,
logistic regression performed well. In most cases, decision trees should perform better in
business scenarios where logistic regression performs better when data contamination is
low. This study can benefit banks by maximizing customer response to product offers,
minimizing overall market costs, and improving customer relationship management |
|---|