Transportation Analytics
Discrete choice models have been used extensively in many areas in economics, marketing, and transportation research. In this competition, we aim to predict the choice among bundles of safety features in cars. The choice-based conjoint data was collected by General Motors to understand consumer’s trade-offs for various vehicle features.
Our team developed multiple models in R such as Multinomial Logit, Mixed Logit, Random Forest, and XGBoost, validating model results on a private test set using a cross-entropy loss metric. We fine-tuned the model with hyperparameter grid search and introduced stacking techniques such as Meta-Learning to implement a soft-voting classifier, obtaining 2nd place overall.
Source Code
tae-hackathon
nathanansel28 • Updated Feb 28, 2025
Report
Alternatively, read the report here: ‣