The 2018 annual Society for Industrial and Organizational Psychology (SIOP) conference featured the first-ever machine learning competition. Teams competed for several months with various analytical techniques to predict turnover in a large US company. Winning submissions had to be done in open source and are posted in this repository. A more complete introduction as presented at the conference can be found here.
- Predict voluntary turnover (who left) vs. who was still active as of December 31, 2014 at Eli Lilly
- Data for 32,296 Eli Lilly employees active as of December 31, 2009
- 162 predictor variables (demographics, location, job-related, job performance, etc.)
- Training set (n = 24,205)
- Test set (n = 8,091)
- Evaluation metric: Cross-validated area under the ROC curve (AUC) statistic.
First Place: An Enriching Meal (Code)
Nick Koenig @ Walmart
David Futrell @ Walmart
Matthew Arsenault @ Walmart
Daniel Schmerling @ Capital One
Private Test Set AUC = .839
Second Place: Team DDI (Code)
Mengqiao (MQ) Liu @ DDI
Rachel King
Evan Sinar
Private Test Set AUC = .836
Third Place: ROC You Like a Hurricane (Code)
Erin Banjanovic @ HumRRO
Adam Beatty @ HumRRO
Ted Diaz @ HumRRO
Rod McCloy @ HumRRO
Colby Nesbitt @ HumRRO
Justin Purl @ HumRRO
Private Test Set AUC = .834
Fourth Place: Byte Monsters (Code)
Isaac Thompson @ Shaker
Scott Tonidandel @ Davidson
Private Test Set AUC = .834
Presentation
Note: this code is split up into smaller chuncks for usability.
Dan J. Putka @ HumRRO
Alexander Schwall
Ben Taylor @ ZIFF