Scalers - Nixtla

`boxcox_lambda`

boxcox_lambda(x, method, season_length=None, lower=-0.9, upper=2.0)

Find optimum lambda for the Box-Cox transformation Parameters:

Name	Type	Description	Default
`x`	`ndarray`	Array with data to transform.	required
`method`	`str`	Method to use. Valid options are ‘guerrero’ and ‘loglik’. ‘guerrero’ minimizes the coefficient of variation for subseries of `x` and supports negative values. ‘loglik’ maximizes the log-likelihood function.	required
`season_length`	`int`	Length of the seasonal period. Only required if method=‘guerrero’.	`None`
`lower`	`float`	Lower bound for the lambda.	`-0.9`
`upper`	`float`	Upper bound for the lambda.	`2.0`

Returns:

Name	Type	Description
`float`	`float`	Optimum lambda.

`boxcox`

boxcox(x, lmbda)

Apply the Box-Cox transformation Parameters:

Name	Type	Description	Default
`x`	`ndarray`	Array with data to transform.	required
`lmbda`	`float`	Lambda value to use.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the transformed data.

`inv_boxcox`

inv_boxcox(x, lmbda)

Invert the Box-Cox transformation Parameters:

Name	Type	Description	Default
`x`	`ndarray`	Array with data to transform.	required
`lmbda`	`float`	Lambda value to use.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the inverted transformation.

`LocalMinMaxScaler`

Bases: _BaseLocalScaler Scale each group to the [0, 1] interval

`LocalMinMaxScaler.fit`

fit(ga)

Compute the statistics for each group. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Name	Type	Description
`self`	`_BaseLocalScaler`	The fitted scaler object.

`LocalMinMaxScaler.fit_transform`

fit_transform(ga)

“Compute the statistics for each group and apply the transformation. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the transformed data.

`LocalMinMaxScaler.inverse_transform`

inverse_transform(ga)

Use the computed statistics to invert the transformation. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the inverted transformation.

`LocalMinMaxScaler.stack`

stack(scalers)

`LocalMinMaxScaler.take`

take(idxs)

`LocalMinMaxScaler.transform`

transform(ga)

Use the computed statistics to apply the transformation. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the transformed data.

`LocalStandardScaler`

Bases: _BaseLocalScaler Scale each group to have zero mean and unit variance

`LocalStandardScaler.fit`

fit(ga)

Compute the statistics for each group. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Name	Type	Description
`self`	`_BaseLocalScaler`	The fitted scaler object.

`LocalStandardScaler.fit_transform`

fit_transform(ga)

“Compute the statistics for each group and apply the transformation. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the transformed data.

`LocalStandardScaler.inverse_transform`

inverse_transform(ga)

Use the computed statistics to invert the transformation. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the inverted transformation.

`LocalStandardScaler.stack`

stack(scalers)

`LocalStandardScaler.take`

take(idxs)

`LocalStandardScaler.transform`

transform(ga)

Use the computed statistics to apply the transformation. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the transformed data.

`LocalRobustScaler`

LocalRobustScaler(scale)

Bases: _BaseLocalScaler Scale each group using robust statistics Parameters:

Name	Type	Description	Default
`scale`	`str`	Type of robust scaling to use. Valid options are ‘iqr’ and ‘mad’. If ‘iqr’ will use the inter quartile range as the scale. If ‘mad’ will use median absolute deviation as the scale.	required

`LocalRobustScaler.fit`

fit(ga)

Compute the statistics for each group. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Name	Type	Description
`self`	`_BaseLocalScaler`	The fitted scaler object.

`LocalRobustScaler.fit_transform`

fit_transform(ga)

“Compute the statistics for each group and apply the transformation. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the transformed data.

`LocalRobustScaler.inverse_transform`

inverse_transform(ga)

Use the computed statistics to invert the transformation. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the inverted transformation.

`LocalRobustScaler.stack`

stack(scalers)

`LocalRobustScaler.take`

take(idxs)

`LocalRobustScaler.transform`

transform(ga)

Use the computed statistics to apply the transformation. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the transformed data.

`LocalBoxCoxScaler`

LocalBoxCoxScaler(method, season_length=None, lower=-0.9, upper=2.0)

Bases: _BaseLocalScaler Find the optimum lambda for the Box-Cox transformation by group and apply it Parameters:

Name	Type	Description	Default
`season_length`	`int`	Length of the seasonal period. Only required if method=‘guerrero’.	`None`
`lower`	`float`	Lower bound for the lambda.	`-0.9`
`upper`	`float`	Upper bound for the lambda.	`2.0`
`method`	`str`	Method to use. Valid options are ‘guerrero’ and ‘loglik’. ‘guerrero’ minimizes the coefficient of variation for subseries of `x` and supports negative values. ‘loglik’ maximizes the log-likelihood function.	required

`LocalBoxCoxScaler.fit`

fit(ga)

Compute the statistics for each group. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Name	Type	Description
`self`	`_BaseLocalScaler`	The fitted scaler object.

`LocalBoxCoxScaler.fit_transform`

fit_transform(ga)

“Compute the statistics for each group and apply the transformation. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the transformed data.

`LocalBoxCoxScaler.inverse_transform`

inverse_transform(ga)

Use the computed lambdas to invert the transformation. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the inverted transformation.

`LocalBoxCoxScaler.stack`

stack(scalers)

`LocalBoxCoxScaler.take`

take(idxs)

`LocalBoxCoxScaler.transform`

transform(ga)

Use the computed lambdas to apply the transformation. Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the transformed data.

`Difference`

Difference(d)

Subtract a lag to each group Parameters:

Name	Type	Description	Default
`d`	`int`	Lag to subtract.	required

`Difference.fit_transform`

fit_transform(ga)

Apply the transformation Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the transformed data.

`Difference.inverse_transform`

inverse_transform(ga)

Invert the transformation Parameters:

Name	Type	Description	Default
`ga`	`GroupedArray`	Array with grouped data.	required

Returns:

Type	Description
`ndarray`	np.ndarray: Array with the inverted transformation.

`Difference.stack`

stack(scalers)

`Difference.take`

take(idxs)

API Reference

​

​boxcox_lambda

​boxcox

​inv_boxcox

​LocalMinMaxScaler

​LocalMinMaxScaler.fit

​LocalMinMaxScaler.fit_transform

​LocalMinMaxScaler.inverse_transform

​LocalMinMaxScaler.stack

​LocalMinMaxScaler.take

​LocalMinMaxScaler.transform

​LocalStandardScaler

​LocalStandardScaler.fit

​LocalStandardScaler.fit_transform

​LocalStandardScaler.inverse_transform

​LocalStandardScaler.stack

​LocalStandardScaler.take

​LocalStandardScaler.transform

​LocalRobustScaler

​LocalRobustScaler.fit

​LocalRobustScaler.fit_transform

​LocalRobustScaler.inverse_transform

​LocalRobustScaler.stack

​LocalRobustScaler.take

​LocalRobustScaler.transform

​LocalBoxCoxScaler

​LocalBoxCoxScaler.fit

​LocalBoxCoxScaler.fit_transform

​LocalBoxCoxScaler.inverse_transform

​LocalBoxCoxScaler.stack

​LocalBoxCoxScaler.take

​LocalBoxCoxScaler.transform

​Difference

​Difference.fit_transform

​Difference.inverse_transform

​Difference.stack

​Difference.take

`boxcox_lambda`

`boxcox`

`inv_boxcox`

`LocalMinMaxScaler`

`LocalMinMaxScaler.fit`

`LocalMinMaxScaler.fit_transform`

`LocalMinMaxScaler.inverse_transform`

`LocalMinMaxScaler.stack`

`LocalMinMaxScaler.take`

`LocalMinMaxScaler.transform`

`LocalStandardScaler`

`LocalStandardScaler.fit`

`LocalStandardScaler.fit_transform`

`LocalStandardScaler.inverse_transform`

`LocalStandardScaler.stack`

`LocalStandardScaler.take`

`LocalStandardScaler.transform`

`LocalRobustScaler`

`LocalRobustScaler.fit`

`LocalRobustScaler.fit_transform`

`LocalRobustScaler.inverse_transform`

`LocalRobustScaler.stack`

`LocalRobustScaler.take`

`LocalRobustScaler.transform`

`LocalBoxCoxScaler`

`LocalBoxCoxScaler.fit`

`LocalBoxCoxScaler.fit_transform`

`LocalBoxCoxScaler.inverse_transform`

`LocalBoxCoxScaler.stack`

`LocalBoxCoxScaler.take`

`LocalBoxCoxScaler.transform`

`Difference`

`Difference.fit_transform`

`Difference.inverse_transform`

`Difference.stack`

`Difference.take`