CodeAlpha_SalesPrediction

Sales prediction using Linear Regression and Random Forest with advertising channel impact analysis.

Sales Prediction using Python 📈

Project Overview

Sales prediction using Linear Regression and Random Forest with advertising channel impact analysis.

Dataset Information

• Total Records: 200
• Source: Kaggle — Advertising Dataset

Features in Dataset

• TV (Advertising Budget in thousands)
• Radio (Advertising Budget in thousands)
• Newspaper (Advertising Budget in thousands)
• Sales (Target Variable in thousands)

Technologies Used

• Python
• Pandas
• NumPy
• Matplotlib
• Seaborn
• Scikit-Learn

Project Workflow

1. Data Loading
2. Data Exploration
3. Data Visualization
4. Train-Test Split
5. Model Training
6. Model Evaluation & Comparison
7. Actual vs Predicted Analysis
8. Channel Impact Analysis
9. Feature Importance Analysis
10. New Sales Prediction
11. Key Insights Summary

Visualizations Created

• Sales Distribution
• Advertising Budget Distribution
• Advertising Channels vs Sales Scatter Plots
• Correlation Heatmap
• Pairplot of All Features
• Actual vs Predicted Comparison Plot
• Channel Impact Bar Chart
• Feature Importance Bar Chart

Machine Learning Models Used

Model 1: Linear Regression Model 2: Random Forest Regressor

• n_estimators = 100
• random_state = 42

Results

Metric	Linear Regression	Random Forest
R² Score	89.94%	98.13%
RMSE	1.78	0.77
MAE	1.46	0.62

Sample Prediction

• TV Budget: $200,000
• Radio Budget: $40,000
• Newspaper Budget: $50,000
• Linear Regression Prediction: 19.63k units
• Random Forest Prediction: 20.26k units

Key Insights

• Random Forest outperformed Linear Regression (98.13% vs 89.94%)
• TV is the most impactful advertising channel on sales
• Newspaper has the least impact on sales
• Average sales across all records is 14.02k units
• Increasing TV budget has the strongest positive effect on sales
• Radio shows moderate correlation with sales

How to Run

1. Clone this repository
2. Install required libraries:

pip install pandas numpy matplotlib seaborn scikit-learn

3. Open the notebook in Jupyter:

jupyter notebook sales_prediction.ipynb

4. Run all cells sequentially

Author

Shreyansh Gupta B.Tech (CSDS) — Galgotias College of Engineering & Technology