This Function is an Implementation of the Holt-Winters' Method for Time Series with Trend and Seasonality. If Necessary it Can Also Return the Best Values for Alpha, Beta and Gama.
-
timeseries = The dataset in a Time Series format.
-
alpha = Level smoothing parameter. The default value is 0.2
-
beta = Trend smoothing parameter. The default value is 0.1
-
gama = Seasonal smoothing parameter. The default value is 0.2
-
m = Seasonal period (Ex: for quarters, m = 4 and for months, m = 12). The default value is 12
-
graph = If True then the original dataset and the moving average curves will be plotted. The default value is True.
-
horizon = Calculates the prediction h steps ahead. The default value is 0.
-
trend = Indicates the types of trend: "additive", "multiplicative" or "none". The default value is "multiplicative".
-
seasonality = Indicates the types of seasonality: "additive", "multiplicative" or "none". The default value is "multiplicative".
Finally a brute force optimization can be done by calling the "optimize_holt" function.