BaseErrorMeasure¶

class pycast.errors.baseerrormeasure.BaseErrorMeasure(minimalErrorCalculationPercentage=60)[source]¶

Bases: pycast.common.pycastobject.PyCastObject

Baseclass for all error measures.

__init__(minimalErrorCalculationPercentage=60)[source]¶

Initializes the error measure.

Parameters:	minimalErrorCalculationPercentage (integer) – The number of entries in an original TimeSeries that have to have corresponding partners in the calculated TimeSeries. Corresponding partners have the same time stamp. Valid values are in [0.0, 100.0].
Raise:	Raises a `ValueError` if minimalErrorCalculationPercentage is not in [0.0, 100.0].

_calculate(startingPercentage, endPercentage, startDate, endDate)[source]¶

This is the error calculation function that gets called by BaseErrorMeasure.get_error().

Both parameters will be correct at this time.

Parameters:	startingPercentage (float) – Defines the start of the interval. This has to be a value in [0.0, 100.0]. It represents the value, where the error calculation should be started. 25.0 for example means that the first 25% of all calculated errors will be ignored. endPercentage (float) – Defines the end of the interval. This has to be a value in [0.0, 100.0]. It represents the value, after which all error values will be ignored. 90.0 for example means that the last 10% of all local errors will be ignored. startDate (float) – Epoch representing the start date used for error calculation. endDate (float) – Epoch representing the end date used in the error calculation.
Returns:	Returns a float representing the error.
Return type:	float
Raise:	Raises a `NotImplementedError` if the child class does not overwrite this method.

_get_error_values(startingPercentage, endPercentage, startDate, endDate)[source]¶

Gets the defined subset of self._errorValues.

Both parameters will be correct at this time.

Parameters:	startingPercentage (float) – Defines the start of the interval. This has to be a value in [0.0, 100.0]. It represents the value, where the error calculation should be started. 25.0 for example means that the first 25% of all calculated errors will be ignored. endPercentage (float) – Defines the end of the interval. This has to be a value in [0.0, 100.0]. It represents the value, after which all error values will be ignored. 90.0 for example means that the last 10% of all local errors will be ignored. startDate (float) – Epoch representing the start date used for error calculation. endDate (float) – Epoch representing the end date used in the error calculation.
Returns:	Returns a list with the defined error values.
Return type:	list
Raise:	Raises a ValueError if startDate or endDate do not represent correct boundaries for error calculation.

confidence_interval(confidenceLevel)[source]¶

Calculates for which value confidenceLevel% of the errors are closer to 0.

Parameters:	confidenceLevel (float) – percentage of the errors that should be smaller than the returned value for overestimations and larger than the returned value for underestimations. confidenceLevel has to be in [0.0, 1.0]
Returns:	return a tuple containing the underestimation and overestimation for the given confidenceLevel
Return type:	tuple
Warning:	Index is still not calculated correctly

get_error(startingPercentage=0.0, endPercentage=100.0, startDate=None, endDate=None)[source]¶

Calculates the error for the given interval (startingPercentage, endPercentage) between the TimeSeries given during BaseErrorMeasure.initialize().

Parameters:	startingPercentage (float) – Defines the start of the interval. This has to be a value in [0.0, 100.0]. It represents the value, where the error calculation should be started. 25.0 for example means that the first 25% of all calculated errors will be ignored. endPercentage (float) – Defines the end of the interval. This has to be a value in [0.0, 100.0]. It represents the value, after which all error values will be ignored. 90.0 for example means that the last 10% of all local errors will be ignored. startDate (float) – Epoch representing the start date used for error calculation. endDate (float) – Epoch representing the end date used in the error calculation.
Returns:	Returns a float representing the error.
Return type:	float
Raise:	Raises a `ValueError` in one of the following cases: startingPercentage not in [0.0, 100.0] endPercentage not in [0.0, 100.0] endPercentage < startingPercentage
Raise:	Raises a `StandardError` if `BaseErrorMeasure.initialize()` was not successfull before.

initialize(*args, **kwargs)[source]¶

Initializes the ErrorMeasure.

During initialization, all BaseErrorMeasure.local_errors() are calculated.

Parameters:	originalTimeSeries (TimeSeries) – TimeSeries containing the original data. calculatedTimeSeries (TimeSeries) – TimeSeries containing calculated data. Calculated data is smoothed or forecasted data.
Returns:	Return `True` if the error could be calculated, `False` otherwise based on the minimalErrorCalculationPercentage.
Return type:	boolean
Raise:	Raises a `StandardError` if the error measure is initialized multiple times.

local_error(originalValue, calculatedValue)[source]¶

Calculates the error between the two given values.

Parameters:	originalValue (list) – List containing the values of the original data. calculatedValue (list) – List containing the values of the calculated TimeSeries that corresponds to originalValue.
Returns:	Returns the error measure of the two given values.
Return type:	numeric
Raise:	Raises a `NotImplementedError` if the child class does not overwrite this method.