Machine learning for pattern discovery in management research. (9th August 2020)

• Record Type: Journal Article
• Authors: Choudhury, Prithwiraj; Allen, Ryan T.; Endres, Michael G.
• Is Part Of: Strategic management journal. Volume 42:Number 1(2021)
• Page Start: 30
• Page End: 57
• Publication Date: 2020-08-09
• Subjects: abduction -- decision trees -- exploratory data analysis -- induction -- machine learning -- neural networks -- partial dependence plots -- pattern discovery -- random forests -- ROC curve -- supervised machine learning; Business planning -- Periodicals; Management -- Periodicals; Business -- Periodicals; 658.401205
• DOI: 10.1002/smj.3215
• Languages: English
• ISSNs: 0143-2095

Abstract

Abstract: Research Summary: Supervised machine learning (ML) methods are a powerful toolkit for discovering robust patterns in quantitative data. The patterns identified by ML could be used for exploratory inductive or abductive research, or for post hoc analysis of regression results to detect patterns that may have gone unnoticed. However, ML models should not be treated as the result of a deductive causal test. To demonstrate the application of ML for pattern discovery, we implement ML algorithms to study employee turnover at a large technology company. We interpret the relationships between variables using partial dependence plots, which uncover surprising nonlinear and interdependent patterns between variables that may have gone unnoticed using traditional methods. To guide readers evaluating ML for pattern discovery, we provide guidance for evaluating model performance, highlight human decisions in the process, and warn of common misinterpretation pitfalls. The Supporting Information section provides code and data to implement the algorithms demonstrated in this article. Managerial Summary: Supervised machine learning (ML) methods are a powerful toolkit that might help managers and researchers discover interesting patterns in large and complex data. We demonstrate this by using several ML algorithms to investigate the drivers of employee turnover at a large technology company. We evaluate the performance of the models, and use visual tools to interpret the patterns revealed. These patterns can be useful in understanding turnover, but we caution not to confuse correlation with causation. These methods should be viewed as "exploratory" and not conclusive proof of relationships in the data. Our guidance can be helpful for managers evaluating analysis conducted by data scientists in their organizations.