Customer Personality Analysis

Overview

Customer Personality Analysis is a detailed examination of a company’s ideal customers. It helps businesses better understand their customers, allowing them to tailor products, marketing strategies, and services to meet the specific needs, behaviors, and concerns of different customer segments.

Instead of marketing a product to all customers, businesses can use customer personality analysis to identify the most relevant customer segment and focus marketing efforts accordingly, improving efficiency and effectiveness.

---

Data Analysis and Preprocessing

The initial dataset contained 2240 samples (rows) and 29 features (columns). The following preprocessing steps were applied:

• Transformed birth date into age (reference date: 01.01.2015) and converted Dt_Customer into days since enrollment.
• Dropped unnecessary columns that did not contribute to meaningful insights.
• Removed invalid marital status entries (YOLO, Absurd, Alone).
• Encoded categorical variables:
  • Label-encoded marital status and education level.
• Handled outliers:
  • Removed outliers in Age and Income using the Interquartile Range (IQR) method.

After cleaning and preprocessing, the final dataset consists of 2198 rows and 20 columns.

---

Key Insights from Correlation Analysis

Income

• Positively correlated with spending on all product categories, especially wine and meat.
• Negatively correlated with the number of children at home.
• High-income customers avoid discounts, indicating lower price sensitivity.

Education

• Higher education levels correlate with increased wine purchases.

Family-Oriented Customers

• Customers with more children at home tend to have higher web visits.
• They exhibit different spending behaviors compared to households without children.

Complaint Behavior & Clustering

• Customers who frequently complain form distinct, small clusters.
• Within each cluster, education levels further segment customer behavior.
• Marital status may not be a significant factor for customer segmentation.
• Customers with children at home ("Kids at Home" and "Teens at Home") are grouped together within specific clusters.

---

Technologies Used

• Python (Pandas, NumPy, Matplotlib, Seaborn)
• Scikit-learn (Dimensionality Reduction & Clustering)
• Data Cleaning & Preprocessing (Feature Engineering, Outlier Removal)
• Visualization (Correlation Heatmaps, Data Distribution Between Clusters)

---

Repository Contents

This repository includes:

• Jupyter Notebook (analysis.ipynb) – Contains data analysis, preprocessing steps, and visualizations.
• Report (report.pdf) – A detailed report summarizing the findings and insights from the analysis.