New in the Second Edition
Readers of the first edition will recognize that this book is substantially longer than that one. That is partly because there are two new chapters—a chapter devoted to treating missing values and another one on pre-processing data for predictive analysis. The insufficient coverage of missing values, and the absence of coverage of pre-processing data for machine learning applications were important omissions. The pre-processing coverage is further improved by new recipes on data pipelines in the final chapter that take the reader from raw data to model evaluation.
The recipes in all chapters have been revised. This is to make sure that they all work well with the most recent versions of pandas. pandas went from version 1.5.3 to 2.2.1 during the writing of this book. I have tried to make sure that all code works fine on all versions of pandas released from January 2023 through February 2024. Since AI tools are becoming increasingly common in our work, I have included discussion of OpenAI tools in four of the chapters. Altogether, 22 of the 82 recipes are new. All of the datasets used have also been updated.
