Linear Regression — E-Commerce Customer Spend Analysis

Repository: Linear Regression Ecommerce
Notebook: Linear Regression Ecommerce.ipynb

🎯 Objective

Predict Yearly Amount Spent by e-commerce customers using behavioral and membership features. Perform exploratory data analysis, fit a regression model, validate results, and provide business recommendations.

---

🔑 Key Results

• Model performance (test set):
  • R² = 0.981
  • RMSE ≈ 10.19
  • MAE ≈ 8.43
• Top drivers of spend:
  • Length of Membership (strongest predictor)
  • Time on App
  • Avg. Session Length
  • Time on Website → negligible impact
• Business implication: Loyalty (membership) and mobile app engagement are the biggest levers to grow revenue.

---

📊 Exploratory Data Analysis

• Pairplot of all variables
  

• Strong linear trend between Length of Membership and Yearly Amount Spent

---

🧑‍💻 Modeling

• Algorithm: Linear Regression (scikit-learn)
• Features used:
  ['Avg. Session Length', 'Time on App', 'Time on Website', 'Length of Membership']
• Target: Yearly Amount Spent
• Train/Test split: 70/30, random_state=42

Evaluation

• Predicted vs Actual values:
  

• Residual distribution:
  

• Q-Q plot for residual normality:
  

---

📈 Business Insights

1. Retention drives revenue → Longer memberships strongly correlate with higher spend.
2. Mobile app is key → App engagement is a bigger driver than website usage.
3. Optimize website funnel → Direct web users toward app for higher conversions.
4. Target high-value users → Use the model to segment and prioritize premium offers.

---

⚠️ Limitations

• Dataset is small (500 rows) — may not generalize.
• Coefficients show correlation, not causation.
• Multicollinearity possible (website/app/session features).

---

▶️ How to Run

1. Clone this repository.
2. Ensure Ecommerce.csv and the notebook are in the root directory.
3. Install dependencies:

pip install pandas matplotlib seaborn scikit-learn scipy jupyter

4. Open and run Linear Regression Ecommerce.ipynb.

---

📂 Project Files

• Linear Regression Ecommerce.ipynb — full analysis notebook
• Dataset/ — Dataset
• images/ — visualizations for README
• Ecommerce.csv — dataset (500 rows)

---

💡Highlights

• Built and validated regression model to predict annual customer spend; achieved R² = 0.981 and RMSE ≈ $10.
• Delivered actionable insights: retention and app engagement as primary drivers of revenue.
• Tools: Python (pandas, scikit-learn, seaborn), EDA, regression modeling, and diagnostics.