On the Epistemology of Data Science: Conceptual Tools for a New Inductivism

This book explores the epistemological foundations of data science, arguing that it is a genuine science that can stand on its own. It proposes an inductivist framework that addresses key concepts such as causation, probability, and analogy and demonstrates its adequacy and usefulness for analyzing the foundations of data science. The book is of interest to computer scientists, philosophers, and data scientists of various disciplines.

Format: Hardback
Length: 295 pages
Publication date: 11 December 2021
Publisher: Springer Nature Switzerland AG

This book delves into the contentious debates surrounding the epistemological foundations of data science. It questions whether data science can be considered a legitimate scientific discipline on its own or if it merely serves as a supplementary practice that can contribute to scientific progress, albeit with limitations. The author presents a comprehensive conceptual framework aimed at addressing these critical questions.

Within the pages of this book, readers will find a robust defense of inductivism, a philosophical approach that emphasizes the importance of drawing conclusions based on empirical evidence. The author also examines the arguments against inductivism and explores the implications of an epistemology of data science that, by definition, must be inductive, given that data science begins with data.

As an alternative to enumerative approaches, the author advocates for a variational rationale in inductive methodology. This rationale suggests that data science can benefit from incorporating various methods and perspectives, rather than relying solely on a single approach. Chapters in the book delve into key concepts of an inductivist methodology, such as causation, probability, and analogy. These concepts are then used to construct an inductivist framework that can be applied to analyze the epistemological foundations of data science.

The author demonstrates that the inductivist framework is both adequate and useful for examining the epistemological underpinnings of data science. It highlights the presence of many aspects of the variational rationale in algorithms commonly used in data science. Additionally, the book includes brief case studies of successful data science applications, such as machine translation, to illustrate the practical applications of the proposed framework.

Data science is positioned within a broader context of several crucial distinctions regarding different types of scientific practices. These distinctions include the distinction between exploratory and theory-driven experimentation and between phenomenological and theoretical science. By exploring these distinctions, the book provides a philosophical perspective that can inform the development and application of algorithms in data science.

This book appeals to a wide range of readers, including computer scientists, philosophers, and data scientists from various disciplines. It serves as a valuable resource for those seeking to deepen their understanding of the epistemological foundations of data science. The conceptual framework presented in the book provides a solid foundation for further in-depth analysis of algorithms used in data science, making it an essential read for anyone interested in this field.

Weight: 641g
Dimension: 235 x 155 (mm)
ISBN-13: 9783030864415
Edition number: 1st ed. 2022