2020 MEBDI ML Competition

Written By Fatih Guvenen

The first annual MEBDI Machine Learning Competition was held in 2020.

The objective of the 2020 Competition was to devise a machine learning algorithm with the best out-of-sample performance for predicting the annual earnings of workers in a sample drawn from the 2012 US Current Population Survey. Further details and competition rules can be found here:  2020 MEBDI ML Competition Details

Submissions were judged by a two-judge faculty panel with one internal member (Professor Joe Mullins) and one outside member (Professor Elena Manresa, NYU Econ), who reviewed the submissions and ran the source codes to verify the results. We are very grateful to Joe and Elena for their invaluable help.

This year’s prizes were funded by Litigation Analytics, Inc., whose support is gratefully acknowledged.

Out of sample fit of the winning ML algorithm compared to a highly flexible Mincer regression with dummies

Winners

Part II (Grand Prize, $4,000)

Awarded to the team of:
Dhananjay Ghei & Sang Min Lee

Read the Executive Summary of their
submission:

Part I: $1000

Equally shared between three teams:

  • Lejvi Dalani and Hasan Tosun

  • Egor Malkov and Filip Premik

  • Dhananjay Ghei and Sang Min Lee

Read the Judging Committee’s Summary:

Fatih Guvenen
Previous

2022 MEBDI ML Competition