2021 Travelers Model Competition
rstats
an insurance fraud detection
Cooperated with two teammates as Team LM, and won the campus winner in UConn. The score of our model ranked first in all 22 teams that competed (final private leaderboard).
Constructed an ensemble boosting model consisting of XGBoost, LightGBM, and AdaBoost, did a thorough data preprocessing, derived some new features from the data, and analyzed the feature importance using LIME.
Gave the final presentation for our team on the Job Shadow Day.