I am trying to create a CV using leave-one-out i.e predicting the next rolling day in my test-data.

I have a train data X_train and test data X_test, and fitting a model using

model.fit(pd.concat((X_train,X_test)))

from fbprophet.diagnostics import cross_validation
cutoffs = X_test["ds"].values
df_cv = cross_validation(model,cutoffs=cutoffs,horizon = '1 day')

th

Is your feature request related to a problem? Please describe.

• Time series forest

Describe the solution you'd like

• Replace loops with joblib parallelized computations or numba.

Description

Now that we have distributions like the Beta distribution, it only makes sense to create a Min-Max Scaler to map the input into the right input domain (support) for the distributions.

It would need to ne something like https://gluon-ts.mxnet.io/api/gluonts/gluonts.block.scaler.html?highlight=meanscaler#gluonts.block.scaler.MeanScaler the MeanScaler.

Describe proposed solution
New function in TimeSeries class named transform taking as parameter function and applying it to TimeSeries itself. For reference see pandas-pipe

signature:
transform(self, function, *args, **kwargs) -> TimeSeries