Closed dvquy13 closed 4 years ago
Google Cloud AutoML with data_fe1:
data_fe1
Wrt accuracy, this model is worse than baseline_0 of 55%.
baseline_0
Google Cloud AutoML with data_fe2 adding basic statistics for each user with each merchant and category and one hot encoding age
data_fe2
merchant
category
age
Adding categorical features in this way actually decrease model performance.
Google Cloud AutoML with
data_fe1
:Wrt accuracy, this model is worse than
baseline_0
of 55%.