Training error for "LINEAR" model

[whoisroop]

Pipeline Configuration:

pj = PredictionJobDataClass(
id=102,
model='linear',
quantiles=[0.1,0.3,0.5,0.7,0.9],
forecast_type="load",
lat=19.0760,
lon=72.8777,
horizon_minutes=24*60,
resolution_minutes=60,
name="Mumbai",
# hyper_params={},
# feature_names=None,
default_modelspecs=None,
save_train_forecasts=True,
)

Training Forecast:

start = time.time()
train_model_pipeline(
pj,
train_data,
check_old_model_age=False,
mlflow_tracking_uri="./mlflow_trained_models",
artifact_folder="./mlflow_artifacts",
)
end = time.time()

ERROR:

ValueError: Input X contains NaN.
LinearRegression does not accept missing values encoded as NaN natively. For supervised learning, you might want to consider sklearn.ensemble.HistGradientBoostingClassifier and Regressor which accept missing values encoded as NaNs natively. Alternatively, it is possible to preprocess the data, for instance by using an imputer transformer in a pipeline or drop samples with missing values. See https://scikit-learn.org/stable/modules/impute.html You can find a list of all estimators that handle NaN values at the following page: https://scikit-learn.org/stable/modules/impute.html#estimators-that-handle-nan-values

[bartpleiter]

Hi Roop,
Thank you for writing this issue. The decision tree based models are able to work with NaNs natively. I would suggest you use xgboost or another GBDT model if possible. If not, then I would happily review your PR to fix the issue! I think adding an optional preprocessing step to input or drop data would be useful!