Train in Data Courses

Master Hyperparameter Optimization for Tabular Learning

Optimize tabular models with cross-validation, grid and random search, Bayesian and multi-fidelity optimization, and Optuna for reliable, efficient tuning.

• Train in Data
• Self-Paced
• Paid Course

Clustering and Dimensionality Reduction

Implement k-means, PCA, UMAP, DBSCAN/HDBSCAN and Louvain clustering from scratch and with scikit-learn, then apply them to single-cell RNA, geospatial and mixed-type data.

• Train in Data
• Self-Paced
• Paid Course

Forecasting with Machine Learning

Forecast single and multiple time series with linear regression, random forests and xgboost: engineer lag and window features, apply recursive and direct multistep strategies, and backtest with error metrics.

• Train in Data
• Self-Paced
• Paid Course

Feature Engineering for Time Series Forecasting

Create lag, window, trend and seasonality features from time series — Fourier terms, calendar holidays, encodings — then forecast with regression models using direct and recursive strategies.

• Train in Data
• Self-Paced
• Paid Course

Feature Engineering for Machine Learning

Transform raw data into model-ready features by imputing missing values, encoding categorical variables, discretizing, scaling, and assembling pipelines with pandas, scikit-learn, and Feature-engine.

• Train in Data
• Self-Paced
• Paid Course

Feature Selection for Machine Learning

Select the best features in a dataset using filter, wrapper, embedded and hybrid methods: correlation, chi-square, ANOVA, Lasso, tree importance and recursive elimination.

• Train in Data
• Self-Paced
• Paid Course

Machine Learning Interpretability

Interpret linear regression, decision trees, random forests, and black-box models in Python using LIME, SHAP, partial dependence, ALE plots, and permutation feature importance.

• Train in Data
• Self-Paced
• Paid Course

Machine Learning with Imbalanced Data

Improve classifiers trained on imbalanced data: apply under-sampling, SMOTE and other over-sampling methods, cost-sensitive learning, ensembles, and metrics like ROC-AUC and balanced accuracy.

• Train in Data
• Self-Paced
• Paid Course