Analyzing Customer Spending Based on Transactional Data Using the Random Forest Algorithm

Abstract

This study explores customer spending behavior using transactional data from a retail dataset, employing a Random Forest Regressor to predict the total amount spent by customers. The dataset includes various customer attributes such as age, gender, and product category, alongside transactional details including quantity purchased and price per unit. Through Exploratory Data Analysis (EDA), it was found that Price and Quantity were the most significant factors influencing total spending, with other features like Age, Gender, and Product Category playing a minimal role in predicting spending behavior. The model achieved perfect accuracy, with an R-squared value of 1.000, indicating that it explained all the variance in customer spending. The findings suggest that transactional features, particularly Price and Quantity, are the key drivers of customer spending, and retailers can optimize their marketing and sales strategies by focusing on these factors. This study also highlights the importance of data preprocessing and feature engineering in enhancing model performance, though the results were limited by the lack of external and behavioral features. Future research could further explore the impact of customer loyalty, external factors, and more complex algorithms to improve predictive accuracy.