Predictive Modelling for Customer Purchase Behaviour: A Logistic Regression Approach Based on Age and Estimated Salary

Customer purchase prediction has become a critical requirement in the insurance industry, where businesses strive to maximize customer acquisition while minimizing marketing costs. Accurate forecasting of whether a potential customer will purchase an insurance policy allows companies to focus on high potential leads and optimize their strategies. In this study, we propose a predictive modelling approach using logistic regression to classify customers based on two key demographic features: Age and Estimated Salary. A dataset of over 1,000 customer records was pre-processed, visualized, and divided into training and testing subsets using an 80:20 ratio. The logistic regression model was trained to identify significant patterns influencing purchase decisions and to estimate the probability of policy adoption. To enhance usability, the trained model was deployed in a Streamlit based web application that includes secure user authentication, interactive input fields, decision boundary visualization, and a leaderboard to track predictive outcomes. Experimental results demonstrate that the logistic regression model achieves an accuracy of approximately 90%, with strong interpretability through coefficient analysis and decision boundary visualization. This work highlights the potential of combining machine learning models with lightweight, interactive applications to support business analysts and decision-makers. The proposed framework offers a scalable, interpretable, and cost-effective solution for insurance companies seeking to strengthen customer targeting. Future work will focus on incorporating additional demographic and behavioral features, applying advanced ensemble models, and integrating large-scale realworld datasets tofurther enhance prediction performance.