Book Image

Data Science Projects with Python - Second Edition

By : Stephen Klosterman
Book Image

Data Science Projects with Python - Second Edition

By: Stephen Klosterman

Overview of this book

If data is the new oil, then machine learning is the drill. As companies gain access to ever-increasing quantities of raw data, the ability to deliver state-of-the-art predictive models that support business decision-making becomes more and more valuable. In this book, you’ll work on an end-to-end project based around a realistic data set and split up into bite-sized practical exercises. This creates a case-study approach that simulates the working conditions you’ll experience in real-world data science projects. You’ll learn how to use key Python packages, including pandas, Matplotlib, and scikit-learn, and master the process of data exploration and data processing, before moving on to fitting, evaluating, and tuning algorithms such as regularized logistic regression and random forest. Now in its second edition, this book will take you through the end-to-end process of exploring data and delivering machine learning models. Updated for 2021, this edition includes brand new content on XGBoost, SHAP values, algorithmic fairness, and the ethical concerns of deploying a model in the real world. By the end of this data science book, you’ll have the skills, understanding, and confidence to build your own machine learning models and gain insights from real data.
Table of Contents (9 chapters)

XGBoost Hyperparameters

Early Stopping

When training ensembles of decision trees with XGBoost, there are many options available for reducing overfitting and leveraging the bias-variance trade-off. Early stopping is a simple one of these and can help provide an automated answer to the question "How many boosting rounds are needed?". It's important to note that early stopping relies on having a separate validation set of data, aside from the training set. However, this validation set will actually be used during the model training process, so it does not qualify as "unseen" data that was held out from model training, similar to how we used validation sets in cross-validation to select model hyperparameters in Chapter 4, The Bias-Variance Trade-Off.

When XGBoost is training successive decision trees to reduce error on the training set, it's possible that adding more and more trees to the ensemble will provide increasingly better fits to the training...