AdRevenue-AI is a machine learning project that predicts product sales based on advertising expenditures across TV, Radio, and Newspapers. The project leverages linear regression to analyze advertising impact and optimize marketing strategies.
- Predict product sales using advertising data.
- Identify the most impactful advertising medium.
- Improve model performance through normalization.
- Visualize insights into advertising effectiveness.
The dataset includes 200 records with the following fields:
- Campaign: Identifier for each advertising campaign.
- TV: Advertising expenditure on TV.
- Radio: Advertising expenditure on Radio.
- Newspaper: Advertising expenditure on Newspapers.
- Sales: Units sold corresponding to advertising.
Key insights:
- TV ads have the highest impact on sales, with the highest R² score.
- Radio ads show moderate correlation with sales (Pearson: 0.35, Spearman: 0.34).
-
Exploratory Data Analysis:
- Imputed missing values for Radio expenditures with the column mean.
- Visualized advertising expenditures and their impact on sales.
-
Model Training:
- Linear regression achieved R² = 0.909 and Adjusted R² = 0.901.
- Normalization improved model performance slightly.
-
Prediction:
- Sales predicted for a new advertising budget (TV=$200, Radio=$40, Newspaper=$50): ~20.5 units.
- Python: Programming language
- Libraries: Pandas, NumPy, Matplotlib, Seaborn, Scikit-learn
- Tools: Jupyter Notebook
- TV advertising is the most significant predictor of sales, with a strong R² score of 0.909.
- Normalization slightly improved the model's performance, aligning feature scales.
- Excluding TV from predictors dropped the R² score to 0.119, showing its critical role.
- Predicting sales for new advertising budgets (TV=$200, Radio=$40, Newspaper=$50) gave ~20.5 units.
- Additional Advertising Platforms: Incorporate digital advertising data for a holistic analysis.
- Advanced Machine Learning Models: Experiment with Ridge Regression, Random Forests, or Neural Networks for better accuracy.
- Web Interface: Build an interactive web app for real-time sales predictions.
- Cost Optimization Analysis: Recommend optimal budgets for maximum sales impact.