sftwre / glm_api Goto Github PK

The API must accept 1 to N rows of model parameters and return the following variables in JSON format:
business outcome - predicted class
phat - predicted probability
params - input parameters

The route for the api should be /predict using POST.

Validate and sanitize input to ensure the model receives correct data.

When np.NaN are encountered in API calls with a single row, mean imputation will drop the column. We need to maintain the columns means on the training data to properly fill empty features.

Ensure API can handle requests with empty, incomplete, and incorrect data.