kats.models.ensemble.weighted_avg_ensemble module

Ensemble models with weighted average individual models

Assume we have k base models, after we make forecasts with each individual model, we learn the weights for each individual model based on corresponding back testing results, i.e., model with better performance should have higher weight.

class kats.models.ensemble.weighted_avg_ensemble.WeightedAvgEnsemble(data: kats.consts.TimeSeriesData, params: kats.models.ensemble.ensemble.EnsembleParams)

Bases: kats.models.ensemble.ensemble.BaseEnsemble

Weighted average ensemble model class

data

the input time series data as in kats.consts.TimeSeriesData

params

the model parameter class in Kats

plot()

Plot method for weighted average ensemble model

predict(steps: int, **kwargs)

Predict method of weighted average ensemble model

Parameters

steps – the length of forecasting horizon

Returns

forecasting results as in pd.DataFrame