Mastering the Machine Learning (ML) Toolkit: Training, Tuning, Evaluating, and Interpreting Predictive Models with Python

This 4-hour, hands-on workshop presented by Ryan Paul Lafler provides a practical introduction to building, training, tuning, and interpreting supervised machine learning (ML) models for predictive analytics. Designed for data scientists, ML engineers, programmers, statisticians, and researchers, it focuses on applying open-source Python libraries to develop predictive models for classification and regression tasks using real-world data.

Through guided exercises, participants will learn how to prepare datasets, conduct Exploratory Data Analysis (EDA), build reproducible ML pipelines, and evaluate model performance using the scikit-learn (sklearn) ecosystem. The workshop emphasizes balancing model complexity and interpretability, addressing overfitting and underfitting, and applying effective feature selection and hyperparameter tuning techniques.

Key topics include:

• Data cleaning and exploratory data analysis (EDA) to uncover feature relationships

• Building end-to-end ML pipelines in scikit-learn

• Training and interpreting supervised models, including LASSO, decision trees, random forests, and gradient-boosted ensembles (AdaBoost, XGBoost)

• Hyperparameter tuning, search spaces, and strategies for improving generalization to unseen data

• Model evaluation techniques, partition strategies, and metric diagnostics for regression and classification

• Understanding the bias-variance tradeoff and model interpretability

All attendees will receive a Certificate of Completion, the PDF slides, a fully documented Jupyter Notebook, and the workshop dataset, equipping them with the tools and confidence to continue training, tuning, and evaluating predictive models in Python. Core libraries include scikit-learn, pandas, numpy, scipy, matplotlib, and seaborn.