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

Preface
Synopsis
Contents
Chapter 1: Introduction
	1.1 Big Data, Machine Learning, and Data Science
	1.2 Ten Theses on Data Science
	References
Chapter 2: Inductivism
	2.1 Characteristics of Inductivism
	2.2 Data Science as an Inductivist Framework
	2.3 An Example from Data Science
	2.4 The Recent Emergence of Inductivist Paradigms in Statistics
	2.5 Arguments Against Inductivism
	References
Chapter 3: Phenomenological Science
	3.1 Theoretical Versus Phenomenological Science
	3.2 Exploratory Experimentation
	3.3 Data Science as Phenomenological Science
	3.4 Truth in Phenomenological Science-Resolving Duhem´s Challenge to Inductivism
	3.5 Case Study: Hierarchies in Deep Neural Networks
	References
Chapter 4: Variational Induction
	4.1 Enumerative, Eliminative, and Variational Induction
	4.2 A Brief History of Variational Induction
		4.2.1 Bacon and His Predecessors
		4.2.2 Herschel and Mill
		4.2.3 Problems of Mill´s Account of Variational Induction
		4.2.4 Twentieth Century Attempts to Solve These Problems
	4.3 Variational Induction in Machine Learning Algorithms
		4.3.1 Decision Trees
		4.3.2 Naïve Bayes
	4.4 Solomonoff Induction
	4.5 Computational Learning Theory
	References
Chapter 5: Causation as Difference Making
	5.1 Causation in Data Science
		5.1.1 The Argument for Causation
		5.1.2 Causation in Data Science
		5.1.3 A Brief Overview of What Is to Come
	5.2 Defining Causation
		5.2.1 Causal Relevance and Causal Irrelevance
		5.2.2 A Difference-Making Account of Causal Counterfactuals
		5.2.3 Remarks on the Asymmetry of Causation
	5.3 A Causal Calculus
		5.3.1 Causal Factors and Alternative Causes
		5.3.2 Effect Factors and Alternative Effects
		5.3.3 Causal Hierarchies and Transitivity
		5.3.4 Functional Dependencies
	5.4 Objections and Criticism
		5.4.1 Objections to Mill´s Method of Difference
		5.4.2 Objections Against Counterfactual Accounts
		5.4.3 Mechanisms and Interventions
		5.4.4 Objections Specific to the Difference-Making Account
		5.4.5 Further Scenarios
	Appendix: The Formal Framework
		A.1 Notation
		A.2 Basic Definitions
			A.2.1 Causal Relevance
				A.2.1.1 Causal Relevance of A  X
				A.2.1.2 Causal Relevance of A  X
			A.2.2 Causal Irrelevance
			A.2.3 Causal Factor
			A.2.4 Alternative Cause
			A.2.5 Inus-Complex
			A.2.6 Homogeneity
				A.2.6.1 Causal Relevance
				A.2.6.2 Causal Irrelevance
			A.2.7 Relevance in Conjunction
		A.3 Simple Theorems
			A.3.1 Lemma 1
			A.3.2 Lemma 2
			A.3.3 Lemma 3
			A.3.4 Lemma 4
			A.3.5 Lemma 5
			A.3.6 Theorem 1
			A.3.7 Theorem 2
			A.3.8 Theorem 3
			A.3.9 Theorem 4
			A.3.10 Theorem 5
			A.3.11 Transitivity
	References
Chapter 6: Evidence
	6.1 Mill´s Methods Revisited
	6.2 Hume´s Problem of Induction Revisited
	6.3 Data Curation and Theory-Ladenness
	References
Chapter 7: Concept Formation
	7.1 Phenomenological and Theoretical Concepts
	7.2 Concept Formation in Phenomenological Science
	7.3 The Problem of Grue
	7.4 Discussion: Supervised and Unsupervised Machine Learning
	References
Chapter 8: Analogy
	8.1 Analogy and Data Science
		8.1.1 Analogical Inferences in Data Science and Beyond
		8.1.2 A Brief Overview of What is to Come
	8.2 Three Historical Perspectives
		8.2.1 Carnap and the Inadequacy of Enumerative Approaches
		8.2.2 Keynes and the Ubiquity of Analogical Reasoning
		8.2.3 Hesse, Bartha and the Two-Dimensional Approach
	8.3 Predictive and Conceptual Analogies
	8.4 A Deterministic Framework for Predictive Analogies
		8.4.1 A First Suggestion
		8.4.2 The Notion of Causal Irrelevance
		8.4.3 A Necessary and Sufficient Criterion
		8.4.4 Conceptual Derivation
		8.4.5 Further Discussion
	8.5 Analogy and Probability
	References
Chapter 9: Causal Probability
	9.1 Probability in Data Science
		9.1.1 The Debate on Probability in Light of Data Science
		9.1.2 A Brief Overview of What Is to Come
	9.2 Predecessors and Contemporary Debate
		9.2.1 Historical Proponents: Cournot, Mill, von Kries
		9.2.2 Contemporary Debate: Abrams, Rosenthal, Strevens
		9.2.3 Some Praise
		9.2.4 Critical Reflections
	9.3 Induction, Causation, and Probability
		9.3.1 Enumerative Induction and the Frequency Theory
		9.3.2 Variational Induction and the Causal Conception of Probability
		9.3.3 A Brief Comparison with Other Accounts
	9.4 Causal Symmetries and the Principle of Causal Symmetry
		9.4.1 Causal Symmetries
		9.4.2 Further Examples
		9.4.3 The Principle of Causal Symmetry
	9.5 Causal Irrelevance and Probabilistic Independence
		9.5.1 Independence
		9.5.2 Interpreting the Measure
	9.6 Ontic and Epistemic Probabilities
		9.6.1 Single-Case Probabilities and Indeterminism
		9.6.2 Epistemic and Ontic Probabilities
		9.6.3 Probabilities from Causal Symptoms
		9.6.4 Probabilities of Causal Hypotheses
	9.7 Conclusion
	References
Chapter 10: Conclusion
Index