DA-Brazilian-ECommerce-Customer-Segmentation

This project analyzes the Brazilian E-Commerce Public Dataset (Olist) to gain insights into customer behavior, purchasing trends, delivery efficiency, and payment methods. The goal is to explore the data, clean and preprocess it, perform analysis, and segment customers based on their purchase patterns.

Dataset Description

The dataset includes multiple related tables covering orders, payments, products, sellers, and customer information.

Key columns used in this analysis:

order_id: Unique order identifier

customer_id: Unique customer identifier

order_status: Status of the order (delivered, shipped, canceled, etc.)

order_purchase_timestamp: Time of purchase

payment_type: Method of payment (credit card, boleto, etc.)

price / freight_value: Product and shipping costs

product_category_name: Product category

customer_city / state: Customer location

️ Steps and Methods

1. Data Cleaning and Preprocessing

Handled missing values appropriately

Converted timestamp columns to datetime format

Corrected data types for numerical and categorical features

Removed duplicates and irrelevant records

2. Feature Engineering

Calculated delivery time, estimated vs. actual delivery difference

Created revenue-related metrics

Extracted customer-level aggregates for segmentation

3. Exploratory Data Analysis (EDA)

Visualized sales over time

Analyzed top-selling categories

Compared actual vs. estimated delivery times

Studied payment distribution and customer states

4. Customer Segmentation

Clustered customers based on purchase frequency, monetary value, and recency

Identified high-value and low-engagement segments

Tools and Technologies

Python

Pandas, NumPy

Matplotlib, Seaborn, Plotly

Scikit-learn (for segmentation)

Jupyter Notebook

? Results and Insights

Most customers preferred credit card payments

Delivery delays were observed in specific regions

Identified top product categories contributing to revenue

Customer segmentation revealed clear behavioral patterns useful for targeted marketing

Future Improvements

Integrate machine learning models for customer lifetime value prediction

Build a dashboard using Power BI or Streamlit

Automate data pipelines for regular updates