{ "version": 13, "cards": [ { "id": "a98ff85a-b8c", "section": "Models", "model": "Linear Regression (OLS)", "tags": [ "Causal", "Linear", "Regression", "Supervised" ], "how_it_works": "- Fits coefficients β to minimize the sum of squared errors\n- β̂ = (XᵀX)⁻¹Xᵀy\n- Assumptions:\n Linearity\n Independent observations\n Exogeneity\n Homoscedasticity\n No multicollinearity\n Normality of the error term", "use_cases": "- Strong baseline for continuous targets\n- Interpretable relationships / marginal effects\n- Forecasting & trend modeling (with proper features)\n- When you want simple, fast, and explainable predictions\n- Establishing linear weights", "link": "https://www.youtube.com/watch?v=NzuK4iAfxhU", "author": "Ben Resnic" }, { "id": "60ba6178-6ea", "section": "Models", "model": "K-Means Clustering", "tags": [ "Clustering", "Unsupervised" ], "how_it_works": "- Groups data points into clusters based on distance from center of clusters\n- Randomly assigns data points as centroids then iterates through all possible centroids to optimize minimum distance\n- Use elbow method or silhouette method to determine optimal number of clusters", "use_cases": "- Separating player archetypes\n- Team style profiles\n- Basically anything where you need to define unique groups based on data", "link": "https://www.youtube.com/watch?v=4b5d3muPQmA", "author": "Jameson Bodenburg" } ] }