Interviewer:
How would you build an ETA model?
আপনি শুরু করতে পারেন:
First I would define exactly what ETA means and at what point the prediction is generated. For example, are we predicting order-placed-to-delivered time or rider-pickup-to-customer time? Then I would identify which features are actually available at that prediction moment to avoid leakage. After that I would establish a simple historical or map-based baseline before trying complex ML models.
এই answer:
“Use XGBoost”
এর চেয়ে অনেক বেশি mature।