/courses/machine-learning/model-optimization/Recent content in Model Optimization & Feature Engineering on DATATWEETSHugoenCopyright (c) 2025 DatatweetsFri, 14 Nov 2025 09:00:00 +0200/courses/machine-learning/model-optimization/lesson-1-feature-engineering/Fri, 14 Nov 2025 09:00:00 +0200/courses/machine-learning/model-optimization/lesson-1-feature-engineering/Discover how feature engineering turns raw columns into better inputs for a model. You will one-hot encode a categorical feature, build informative ratio features, and tame a skewed feature with a log transform on the real California Housing dataset, measuring the honest impact on model performance./courses/machine-learning/model-optimization/lesson-2-model-selection/Fri, 14 Nov 2025 09:00:00 +0200/courses/machine-learning/model-optimization/lesson-2-model-selection/Learn how to put several machine learning algorithms on a level playing field, compare their test performance on the real California Housing dataset, and reason about the trade-offs between accuracy, interpretability, and speed when selecting a model./courses/machine-learning/model-optimization/lesson-3-cross-validation/Fri, 14 Nov 2025 09:00:00 +0200/courses/machine-learning/model-optimization/lesson-3-cross-validation/Discover why one train/test split can mislead you, then learn how k-fold cross-validation produces a steadier estimate of how a model performs on unseen data. You will use KFold and cross_val_score on the real California Housing dataset, interpret the mean and spread of fold scores, and reason about the bias-variance trade-off when choosing the number of folds./courses/machine-learning/model-optimization/lesson-4-regularization/Fri, 14 Nov 2025 09:00:00 +0200/courses/machine-learning/model-optimization/lesson-4-regularization/Learn how regularization simplifies models so they generalize better. You will add a penalty term to linear regression, tune the strength parameter alpha with cross-validation, and compare Ridge (L2) and Lasso (L1) on the real California Housing dataset./courses/machine-learning/model-optimization/lesson-5-going-beyond-linear-models/Fri, 14 Nov 2025 09:00:00 +0200/courses/machine-learning/model-optimization/lesson-5-going-beyond-linear-models/Linear models only draw straight lines. In this lesson you will give them curves with polynomial features, watch underfitting and overfitting play out as you raise the degree, and then switch to tree-based models that learn nonlinear patterns on their own. You will compare linear regression against decision trees, random forests, and gradient boosting on the real California Housing dataset./courses/machine-learning/model-optimization/lesson-6-guided-project-optimizing-model-prediction/Fri, 14 Nov 2025 09:00:00 +0200/courses/machine-learning/model-optimization/lesson-6-guided-project-optimizing-model-prediction/Combine everything from this module in one guided project: engineer features, log-transform a skewed target, compare four models with cross-validation, and apply regularization on the real Forest Fires dataset. You will discover that all models score near or below zero R2, and learn why that honest finding is worth more than a fabricated success.