Neal Lathia
Senior Data Scientist at Skyscanner (London). Applied Learning to Rank to flight itinerary search, demonstrating the full LTR workflow from feature engineering to A/B testing in a travel search context.
Articles
- Learning to Rank for Flight Itinerary Search — LTR for flights, purchase-completion labels, MAP/MRR evaluation
Key Contributions
- Showed that purchase completion outperforms clicks as a relevance signal for flight ranking
- Demonstrated offline MAP/MRR predicted online conversion improvement
- Simple logistic regression LTR outperformed rule-based flight ranking
Related
- Skyscanner · Learning to Rank · MRR · MAP