Functions used to create pytorch `DataSet`s and `DataLoader`s.

convert_date[source]

convert_date(dates)

Converts pandas date format to torch Tensors (floats)

get_moments[source]

get_moments(df)

Gets the mean and variance of all columns.

create_tensors[source]

create_tensors(df, moments, predict=False)

converts a pandas dataframe to pytorch tensors

class TimeSeries[source]

TimeSeries(x, y) :: Dataset

Load raw x,y data

class DataFrame[source]

DataFrame(df, moments=None) :: Dataset

Take a data frame with columns ds | y | other_cols and convert to pytorch tensors.

Functions to create Fastai DataBunches.

create_db[source]

create_db(df, train_p=0.8, bs=96, moments=None)

Take dataframe and convert to Fastai databunch