This documentation is built from the development branch, master. It includes the changes listed under Unreleased, which the latest release, CityLearn 3.6.0, does not have. To read the documentation of the version you installed (pip show citylearn), open the docs folder at its tag on GitHub, for example v3.6.0.

citylearn.preprocessing module

class citylearn.preprocessing.Encoder[source]

Bases: object

Base class to transform observations.

class citylearn.preprocessing.NoNormalization[source]

Bases: Encoder

Use to return observation value as-is i.e. without any transformation.

Examples

>>> x_max = 24
>>> encoder = NoNormalization()
>>> observation = 2
>>> encoder*observation
2
class citylearn.preprocessing.Normalize(x_min: float | int, x_max: float | int)[source]

Bases: Encoder

Use to transform observations to a value between x_min and x_max using min-max normalization.

Parameters:
  • x_min (Union[float, int]) – Minimum observation value.

  • x_max (Union[float, int]) – Maximum observation value.

Notes

The transformation returns two values \(x_{sin}\) and \(x_{sin}\) defined as:

\[x = \frac{x - x_{min}}{x_{max} - x_{min}}\]

Examples

>>> x_min = 0
>>> x_max = 24
>>> encoder = Normalize(x_min, x_max)
>>> observation = 2
>>> encoder*observation
0.08333333333333333
class citylearn.preprocessing.OnehotEncoding(classes: List[float] | List[int] | List[str])[source]

Bases: Encoder

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: float | int)[source]

Bases: Encoder

Use to transform observations that are cyclical/periodic e.g. hour-of-day, day-of-week, e.t.c.

Parameters:

x_max (Union[float, int]) – Maximum observation value.

Notes

The transformation returns two values \(x_{sin}\) and \(x_{sin}\) defined as:

\[ \begin{align}\begin{aligned}x_{sin} = sin(\frac{2 \cdot \pi \cdot x}{x_{max}})\\x_{cos} = cos(\frac{2 \cdot \pi \cdot x}{x_{max}})\end{aligned}\end{align} \]

Examples

>>> x_max = 24
>>> encoder = PeriodicNormalization(x_max)
>>> observation = 2
>>> encoder*observation
array([0.75, 0.9330127])
class citylearn.preprocessing.RemoveFeature[source]

Bases: Encoder

Use to exlude an observation by returning None type.

Examples

>>> encoder = RemoveFeature()
>>> observation = 2
>>> encoder*observation
None