Project Overview

Data Overview

The dataset consists of purchase data of individual customers of a clothing retailer, covering:
- Demographic attributes (e.g., age, gender)
- Product details (e.g., category, size, color)
- Purchase location and season
- Transaction data (e.g., payment method, binary indicator for promotion code usage)

We initially explored multiple directions through EDA but didn’t identify any clear or noticeable trends.
Upon inspection, we realized the original dataset was AI-generated (via ChatGPT), which may account for the overly uniform and synthetic distribution. (see chart on the right).

Given the time constraints of the project, we decided not to replace the dataset.
Instead, we divided the workload by field characteristics that allowed for meaningful segmentation (e.g., demographic-focused, transaction-focused).
Each team member explored their respective area through external data enrichment, data visualization, and drill-down analysis.

I focused on EDA by state and season, and observed regional differences in seasonal purchasing patterns, which I found intriguing.
To enrich the analysis, I sourced and integrated U.S. regional climate data and explored seasonality in clothing purchases across different states.
I used Tableau to visualize regional trends and attempted to identify meaningful seasonal signals from a visual perspective.

I aggregated seasonal purchase counts across all 50 U.S. states using a pivot table and visualized the results with a heatmap.
To capture temporal trends more intuitively, I also supplemented the view with a line chart for multi-dimensional analysis.
The visualizations revealed distinct seasonal shopping patterns across states. While some states saw higher purchase volumes during the summer, others peaked in winter. Notably, Vermont exhibited a strong concentration of purchases in the winter season.