Fillers
The cesnet_tszoo
package supports various ways of dealing with missing data in dataset.
Possible config parameters in TimeBasedConfig
and SeriesBasedConfig
:
fill_missing_with
: Can pass enumFillerType
for built-in filler or pass a type of custom filler that must derive fromFiller
base class.default_values
: Default values for missing data, applied before fillers. Can set one value for all features or specify for each feature.
Note
Fillers can carry over values from the train set to the validation and test sets. For example, ForwardFiller
can carry over values from previous sets.
Built-in fillers
The cesnet_tszoo
package comes with multiple built-in fillers. To check built-in fillers refer to fillers
.
Custom fillers
It is possible to create and use own fillers. But custom filler must derive from Filler
base class.