The book is being continuously updated (" similar to how software is developed"), and Molnar invites others to contribute as well. Reading the book is recommended for machine learning practitioners, data scientists, statisticians, and anyone else interested in making machine learning models interpretable. The book focuses on machine learning models for tabular data (also called relational or structured data) and less on computer vision and natural language processing tasks. Most chapters follow a similar structure and focus on one interpretation method.Īnd what type of data being modeled does this book center on? I only recommend that you start with the introduction and the chapter on interpretability. Ou can jump back and forth and concentrate on the techniques that interest you most. If you don't have the time or interest to read from cover to cover, Molnar says: The book's table of contents are as follows: Instead, expect the distilled expertise of someone who is clearly invested in and passionate about the topic, and has been studying it in-depth. Recall that Molnar is a data scientist and PhD candidate in interpretable machine learning, meaning that you should be able to rest assured that this won't be a collection of outdated or marginal ideas related to the subject.
Later chapters focus on general model-agnostic methods for interpreting black box models like feature importance and accumulated local effects and explaining individual predictions with Shapley values and LIME. The book is described as progressing as follows:Īfter exploring the concepts of interpretability, you will learn about simple, interpretable models such as decision trees, decision rules and linear regression. No books, no tutorials, no overview papers, nothing. But I only found the relevant research papers and a few blog posts scattered around the internet, but nothing with a good overview.
Given the success of machine learning and the importance of interpretability, I expected that there would be tons of books and tutorials on this topic. Molnar goes on to say in the book's preface:
This book is about making machine learning models and their decisions interpretable. But computers usually do not explain their predictions which is a barrier to the adoption of machine learning. Machine learning has great potential for improving products, processes and research.
0 kommentar(er)
