© Institute of Mathematical Statistics, 2010 To Explain or to Predict?
By G. Shmueli
2,118 citations recorded by the source corpus.
Statistical modeling is a powerful tool for developing and testing theories by way of causal explanation, prediction, and description. In many disciplines there is near-exclusive use of statistical modeling for causal ex- planation and the assumption that models with high explanatory power are inherently of high predictive power. Conflation between explanation and pre- diction is common, yet the distinction must be understood for progressing scientific knowledge. While this distinction has been recognized in the phi- losophy of science, the statistical literature lacks a thorough discussion of the many differences that arise in the process of modeling for an explanatory ver- sus a predictive goal. The purpose of this article is to clarify the distinction between explanatory and predictive modeling, to discuss its sources, and to reveal the practical implications of the distinction to each step in the model- ing process.
Subfields: Artificial Intelligence, History and Philosophy of Science, Statistics and Probability, Statistical Learning Theory
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Cynthia Rudin
- Christian M. Ringle
- Marko Sarstedt
- Joseph F. Hair
- Jimeng Sun
- Berk Ustun
- Edward Choi
- Margo I. Seltzer
In History and Philosophy of Science
- David Muchlinski
- Frank Emmert-Streib
- Tarin Paz-Kagan
- Sara Leroi-Werelds
- Florian Hartig
- Ewout W. Steyerberg
- David S. Siroky
- Flávio du Pin Calmon
In Statistics and Probability
- Cynthia Rudin
- Christian M. Ringle
- Berk Ustun
- Joseph F. Hair
- Jörg Henseler
- Y. Vergouwe
- Ewout W. Steyerberg
- Margo I. Seltzer
In Statistical Learning Theory
- Berk Ustun
- Margo I. Seltzer
- Mohammed Berrada
- Amina Adadi
- Nicholas Larus-Stone
- Daniel Alabi
- Elaine Angelino
- Hongyu Yang
Citations in the corpus
- Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI) (2018, background): “[165] G. Shmueli, ‘‘To explain or to predict?’’”
- Machine Learning Model Interpretability for Precision Medicine (2016, background): “In the trade-off between predictive modelling and explanatory modelling [2], explanatory modeling is highly important among healthcare practitioners because they value the ability to understand the contribution of specific features to a model.”
- Forest Floor Visualizations of Random Forests (2016, background): “RF can be termed a predictive algorithmic model, designed to have a high predictive accuracy on the expense of model transparency [22, 3].”
- Learning Certifiably Optimal Rule Lists (2017, background): “There is a growing interest in interpretable (transparent, comprehensible) models because of their societal importance (see Rüping, 2006; Bratko, 1997; Dawes, 1979; Vellido et al., 2012; GiraudCarrier, 1998; Holte, 1993; Shmueli, 2010; Huysmans et al., 2011; Freitas, 2014).”
- Predictive Multiplicity in Classification (2019, background): “Many models are now built merely to make accurate predictions [24, 45].”
- Scalable Bayesian Rule Lists (2016, background): “Interpretability has long since been a fundamental topic in artificial intelligence (see R¨uping, 2006; Bratko, 1997; Dawes, 1979; Vellido, Mart´ın-Guerrero, & Lisboa, 2012; Giraud-Carrier, 1998; Holte, 1993; Shmueli, 2010; Huysmans, Dejaeger, Mues, Vanthienen, & Baesens, 2011; Freitas, 2014).”
- Comparing Random Forest with Logistic Regression for Predicting Class-Imbalanced Civil War Onset Data (2016, positive): “The ability of a statistical method to make accurate predictions is just as important as its ability to explain causal processes, and there is enough room for both standards of evaluation within the discipline (Shmueli 2010).”
- Comparing sample bias correction methods for species distribution modeling using virtual species (2021, methodology): “In general, machine learning methods are better suited for prediction and less so at identifying and explaining causal relationships (Shmueli 2010).”
- On the Emancipation of PLS-SEM: A Commentary on Rigdon (2012) (2014, methodology): “For this purpose, a predictive modeling procedure specific to PLS-SEM should be established that highlights the specifics of predictive modeling in terms of model building and assessment (Shmueli, 2010).”
- Mirror, mirror on the wall: a comparative evaluation of composite-based structural equation modeling methods (2017, background): “As such, factor-based SEM focuses on explanatory modeling, which is Bthe use of statistical models for testing causal explanations^ (Shmueli 2010, p. 390).”
- Can we avoid reductionism in risk reduction? (2017, background): “…can have higher predictive validity than a model that correctly specifies the “true” causal relationship between independent and dependent variables (Shmueli, 2010).1 Under certain assumptions, all causal effects will have some predictive value but not all predictive associations will have…”