Forecasting Methods

class pycast.methods.exponentialsmoothing.ExponentialSmoothing(smoothingFactor=0.1, valuesToForecast=1)

Bases: pycast.methods.basemethod.BaseForecastingMethod

Implements an exponential smoothing algorithm.

Explanation:

http://www.youtube.com/watch?v=J4iODLa9hYw

__init__(smoothingFactor=0.1, valuesToForecast=1)

Initializes the ExponentialSmoothing.

Parameters:

• smoothingFactor (float) – Defines the alpha for the ExponentialSmoothing. Valid values are in (0.0, 1.0).
• valuesToForecast (integer) – Number of values that should be forecasted.

Raise:

Raises a ValueError when smoothingFactor has an invalid value.

_get_parameter_intervals()

Returns the intervals for the methods parameter.

Only parameters with defined intervals can be used for optimization!

Returns:Returns a dictionary containing the parameter intervals, using the parameter name as key, while the value hast the following format: [minValue, maxValue, minIntervalClosed, maxIntervalClosed]

• minValue

  Minimal value for the parameter

• maxValue

  Maximal value for the parameter

• minIntervalClosed

  True, if minValue represents a valid value for the parameter. False otherwise.

• maxIntervalClosed:

  True, if maxValue represents a valid value for the parameter. False otherwise.

Return type:dictionary

execute(timeSeries)

Creates a new TimeSeries containing the smoothed and forcasted values.

Returns:TimeSeries object containing the smoothed TimeSeries, including the forecasted values. Return type:TimeSeries Note:The first normalized value is chosen as the starting point.

class pycast.methods.exponentialsmoothing.HoltMethod(smoothingFactor=0.1, trendSmoothingFactor=0.5, valuesToForecast=1)

Bases: pycast.methods.basemethod.BaseForecastingMethod

Implements the Holt algorithm.

Explanation:

http://en.wikipedia.org/wiki/Exponential_smoothing#Double_exponential_smoothing

__init__(smoothingFactor=0.1, trendSmoothingFactor=0.5, valuesToForecast=1)

Initializes the HoltMethod.

Parameters:

• smoothingFactor (float) – Defines the alpha for the ExponentialSmoothing. Valid values are in (0.0, 1.0).
• trendSmoothingFactor (float) – Defines the beta for the HoltMethod. Valid values are in (0.0, 1.0).
• valuesToForecast (integer) – Defines the number of forecasted values that will be part of the result.

Raise:

Raises a ValueError when smoothingFactor or trendSmoothingFactor has an invalid value.

_get_parameter_intervals()

Returns the intervals for the methods parameter.

Only parameters with defined intervals can be used for optimization!

Returns:Returns a dictionary containing the parameter intervals, using the parameter name as key, while the value hast the following format: [minValue, maxValue, minIntervalClosed, maxIntervalClosed]

• minValue

  Minimal value for the parameter

• maxValue

  Maximal value for the parameter

• minIntervalClosed

  True, if minValue represents a valid value for the parameter. False otherwise.

• maxIntervalClosed:

  True, if maxValue represents a valid value for the parameter. False otherwise.

Return type:dictionary

execute(timeSeries)

Creates a new TimeSeries containing the smoothed values.

Returns:TimeSeries object containing the smoothed TimeSeries, including the forecasted values. Return type:TimeSeries Note:The first normalized value is chosen as the starting point.

class pycast.methods.exponentialsmoothing.HoltWintersMethod(smoothingFactor=0.1, trendSmoothingFactor=0.5, seasonSmoothingFactor=0.5, seasonLength=0, valuesToForecast=1)

Bases: pycast.methods.basemethod.BaseForecastingMethod

Implements the Holt-Winters algorithm.

Explanation:

http://en.wikipedia.org/wiki/Exponential_smoothing#Triple_exponential_smoothing

__init__(smoothingFactor=0.1, trendSmoothingFactor=0.5, seasonSmoothingFactor=0.5, seasonLength=0, valuesToForecast=1)

Initializes the HoltWintersMethod.

Parameters:

• smoothingFactor (float) – Defines the alpha for the Holt-Winters algorithm. Valid values are (0.0, 1.0).
• trendSmoothingFactor (float) – Defines the beta for the Holt-Winters algorithm.. Valid values are (0.0, 1.0).
• seasonSmoothingFactor (float) – Defines the gamma for the Holt-Winters algorithm. Valid values are (0.0, 1.0).
• seasonLength (integer) – The expected length for the seasons. Please use a good estimate here!
• valuesToForecast (integer) – Defines the number of forecasted values that will be part of the result.

_get_parameter_intervals()

Returns the intervals for the methods parameter.

Only parameters with defined intervals can be used for optimization!

Returns:Returns a dictionary containing the parameter intervals, using the parameter name as key, while the value hast the following format: [minValue, maxValue, minIntervalClosed, maxIntervalClosed]

• minValue

  Minimal value for the parameter

• maxValue

  Maximal value for the parameter

• minIntervalClosed

  True, if minValue represents a valid value for the parameter. False otherwise.

• maxIntervalClosed:

  True, if maxValue represents a valid value for the parameter. False otherwise.

Return type:dictionary

computeA(j, timeSeries)

Calculates A_j. Aj is the average value of x in the jth cycle of your data

Returns:A_j Return type:numeric

execute(timeSeries)

Creates a new TimeSeries containing the smoothed values.

Returns:TimeSeries object containing the exponentially smoothed TimeSeries, including the forecasted values. Return type:TimeSeries Note:Currently the first normalized value is simply chosen as the starting point.

initSeasonFactors(timeSeries)

Computes the initial season smoothing factors.

Returns:Returns a list of season vectors of length “seasonLength”. Return type:list

initialTrendSmoothingFactors(timeSeries)

Calculate the initial Trend smoothing Factor b0.

Explanation:

http://en.wikipedia.org/wiki/Exponential_smoothing#Triple_exponential_smoothing

Returns:Returns the initial trend smoothing factor b0