mcgill-mma-enterpriseanalytics / customer-personality-analysis

Customer Personality Analysis Project

Overview

This project utilizes data from Kaggle's Customer Personality Analysis to explore customer segmentation and its impact on marketing strategies. Through rigorous preprocessing, including outlier removal, iterative imputation, feature engineering, and one-hot encoding, we delve into clustering methods to affirm our hypothesis:

1. Segmentation enhances promotion effectiveness
2. Customized offers boost customer engagement
3. Data-driven insights lead to increased sales

Methodology

Data Preprocessing: Outlier removal, iterative imputation for missing values, and feature engineering.

Clustering Analysis: Utilized the elbow method for optimal cluster determination, followed by customer segmentation into four personas based on demographic and behavioral insights.

Causal Inference Models: Investigated the effect of marital status, complaints, and income on total purchases using CausalML model.

Findings

We identified four distinct customer segments:

• Savvy Small Couples: High-income, promotion-accepting small couples with moderate web purchases.
• Family Savers: Married with larger families, lower income, moderate deal sensitivity.
• Budget-Conscious Families: Larger families, lower income, small purchase value.
• Market Value Seekers: Moderately higher income, larger family size, and a preference for web purchases.

Marketing Suggestions

Customized marketing strategies were developed for each segment, focusing on premium services, exclusive sales, value bundles, and essential goods discounts to enhance customer engagement and sales.

Model Validity and Future Enhancements

• Models Validity Enhancement: Suggestions include consensus analysis with other algorithms, self-supervised learning, and incorporating R learners for causal inference.
• Data Expansion: Updating datasets, integrating diverse data types, and increasing data volume for robust analysis.

Conclusion

Our analysis demonstrates the power of customer segmentation in crafting targeted marketing strategies.

Future work will focus on enhancing model validity, expanding data sources, and integrating findings into business strategies for targeted engagement and product development.