citylearn.preprocessing module

class citylearn.preprocessing.Encoder

Bases: object

class citylearn.preprocessing.NoNormalization

Bases: citylearn.preprocessing.Encoder

class citylearn.preprocessing.Normalize(x_min: Union[float, int], x_max: Union[float, int])

Bases: citylearn.preprocessing.Encoder

class citylearn.preprocessing.OnehotEncoding(classes: Union[List[float], List[int], List[str]])

Bases: citylearn.preprocessing.Encoder

Initialize PeriodicNormalization encoder class.

Use to transform unordered categorical observations e.g. boolean daylight savings e.t.c.

Parameters

classes (Union[List[float], List[int], List[str]]) – Observation categories.

Examples

>>> classes = [1, 2, 3, 4]
>>> encoder = OnehotEncoding(classes)
# identity_matrix = [[1, 0, 0, 0], [0, 1, 0, 0], [0, 0, 1, 0], [0, 0, 0, 1]]
>>> observation = 2
>>> encoder*observation
[0, 1, 0, 0]

class citylearn.preprocessing.PeriodicNormalization(x_max: Union[float, int])

Bases: citylearn.preprocessing.Encoder

class citylearn.preprocessing.RemoveFeature

Bases: citylearn.preprocessing.Encoder