This project analyzes the personality of customers of a company that sells products through various channels (physical store, website, catalog), using purchase data, campaigns and demographic characteristics.
- Customer Segmentation: Group customers according to purchasing behavior and demographic characteristics.
- Identification of Popular Products: Determine the most purchased products and the segments with the highest spending.
- Campaign Effectiveness Evaluation: Measure the effectiveness of marketing campaigns according to client characteristics.
- Campaign Response Prediction: Predict the probability that a customer will respond to future campaigns.
- Optimization of Marketing Strategies: Focus resources on the most valuable segments.
- Segmentation: Use of the Elbow Method to determine the optimal number of clusters (3 suggested clusters).
- Campaign Effectiveness Analysis: Visualization of response rates, segmentation by demographic characteristics and ROI analysis.
- Predictive Models: Classification with Random Forest to predict the response to campaigns based on income, total spending and in-store purchases.
- Identified Segments: Three main segments:
- High income/expense
- Young people with low spending
- Frequent buyers in stores
- Campaign Response Rate: The overall rate is low (7% max.), but certain segments respond better.
- Model Accuracy: 84% accuracy, although the ROC AUC is low (0.55), which indicates opportunity for improvement.