This is a companion discussion topic for the original entry at https://dphi.tech/challenges/252
Why is the evaluation based on RMSE when the target column is categorical? There are exactly 7 classes for delivery times ranging from 10 minutes to 120 minutes.
Because in real life, time is regression one, it’s not like delivery happen in 5 & 10 mins but in mids tooo
In real life, restaurants show an expected delivery time like 10 mins, 30 mins or 45 mins just like the categories in the dataset. Nobody predicts delivery time will be 23.43 mins. That is why the data has exactly 7 categories of expected delivery time.