Boosting Customer Retention in Pharmaceutical Retail: A Predictive Approach Based on Machine Learning Models

Prediction models have acquired significant importance in business decision-making by allowing the use of data to generate strategic information. This study focuses on creating and evaluating a predictive model to identify possible customer abandonment in a pharmaceutical retail company to promote retention strategies. The approach is based on applying the Cross-Industry Standard Process for Data Mining (CRISP-DM) method and using Python as the primary data science tool. Three classification algorithms were considered to predict customer abandonment: neural networks, decision trees, and logistic regression. These algorithms were trained and evaluated using historical data from a pharmaceutical company. After an exhaustive analysis of the results, we concluded that the neural network model proves the most appropriate, reaching an impressive accuracy of 0.99 in the classification. The implementation of this model will allow the pharmaceutical retail company to identify in advance the customers with the highest risk of abandoning their services. This identification enables the marketing team to implement personalized and timely retention strategies, thus reducing the churn rate and improving the overall efficiency and effectiveness of the organization.