Closed kaylarobinson077 closed 1 year ago
I didn't change the target feature to be between 0-1. Since we decided to do a regression I kept the target as is so it would predict the # of days. I wasn't too sure how we would scale the target to be a % output of adoptability. The model isn't all that accurate and I ended up spending more time trying to retraining the model. In the future if there's a way we can get a binary feature that is [0,1] for adopted vs. not we should consider doing a survival analysis instead of reg.
Standardize LOS output to be a [0,1] as a percentile compared to the training dataset