Taylor Bell

I've actually seen a similar problem, and I am currently preparing an example to share (watch as it works this time!)

Here's a [gist with train and test files](https://gist.github.com/tayiorbeii/975a12a3d45c086d01f80a159095dee6). I ran the following command to target the `result` column: ``` automl_gs train.csv result --model_name broken_example --framework xgboost --num_trials 100 ``` then...

Adding this line before each encoder transform call in `pipeline.py` is helping, but annoying to do manually: `whatever_enc = whatever_enc[~np.isnan(whatever_enc)]` ...but different errors are good to see! EDIT: Found some...


@tingxuanz I think it sounds like a great idea if you are up for it!