Calibration after bootstrap for accurate uncertainty quantification in regression models

• Published in: npj computational materials Vol. 8; no. 1; pp. 1 - 9
• Main Authors: Palmer, Glenn, Du, Siqi, Politowicz, Alexander, Emory, Joshua Paul, Yang, Xiyu, Gautam, Anupraas, Gupta, Grishma, Li, Zhelong, Jacobs, Ryan, Morgan, Dane
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group 20.05.2022; Nature Portfolio
• Subjects: Accuracy; Calibration; Estimates; Learning algorithms; Machine learning; Materials science; Methods; Model accuracy; Neural networks; Regression analysis; Regression models; Standard deviation; Statistical analysis; Statistical methods; Uncertainty
• ISSN: 2057-3960; 2057-3960
• DOI: 10.1038/s41524-022-00794-8

Abstract Obtaining accurate estimates of machine learning model uncertainties on newly predicted data is essential for understanding the accuracy of the model and whether its predictions can be trusted. A common approach to such uncertainty quantification is to estimate the variance from an ensemble of models, which are often generated by the generally applicable bootstrap method. In this work, we demonstrate that the direct bootstrap ensemble standard deviation is not an accurate estimate of uncertainty but that it can be simply calibrated to dramatically improve its accuracy. We demonstrate the effectiveness of this calibration method for both synthetic data and numerous physical datasets from the field of Materials Science and Engineering. The approach is motivated by applications in physical and biological science but is quite general and should be applicable for uncertainty quantification in a wide range of machine learning regression models.