This notebook demonstrates a complete workflow for building, training, and visualizing a multiple linear regression model in Python. It sets up a reproducible environment using common data science libraries, fits a regression model using scikit-learn, and extracts the learned coefficients and intercept for inspection. Custom functions are included to format plots for clarity and reusability across projects. For the accompanying walkthrough video, I anonymized the data to be able to add it to my portfolio.
To visualize the model’s performance, the notebook generates 3D scatter plots from multiple perspectives, showing both the original data and the fitted regression surfaces. These visualizations help reveal how the model responds to variation in the inputs, making its behavior more interpretable. The code is modular, well-commented, and designed to support future experimentation with different modeling and visualization approaches.
The bottom-left plot was the money shot. The regression line fits closely along a visible trajectory through the data cloud, with minimal dispersion, which suggested strong predictive alignment between the inputs (avg games/wk and game completion rate) and the output (average customer lifetime value).