pyhts 0.1.0

A package for forecasting reconciliation.

py-hts

A python package for hierarchical forecasting, inspired by hts package in R.

features

• support pupular forecast reconciliation models in the literature, e.g. ols, wls, mint et al. Temporal Hierarchy will be supported in the future.
• multiple methods for the construction of hierarchy.
• use different base forecasters for different hierarchical levels.
• familiar sklearn-like API

Quick Demo

Load tourism data, which is tourism demand measured by the number of "visitor nights" in Australia.

from pyhts.dataset import load_tourism

tourism_data = load_tourism()
train = tourism_data.iloc[:-12, :]
test = tourism_data.iloc[-12:, :]

Define hierarchy

from pyhts.hierarchy import Hierarchy
hierarchy = Hierarchy.from_names(tourism_data.columns, chars=[1, 1, 1])
print(hierarchy.node_name)

Create an ols forecasting reconciliation model with sklearn-like API.

from pyhts.HFModel import HFModel
model = HFModel(hierarchy=hierarchy, base_forecasters="arima", 
                hf_method="comb", comb_method="ols")

Fit the model and forecast.

model.fit(train)
forecasts = model.predict(horizon=12)

• model.fit() will fit the baseforecasters and compute the weighting matrix used to reconcile the base forecast.

• model.forecast() will calculate base forecasts of all levels and reconcile the base forecasts.

Obtain coherent forecasts of all hierarchical levels.

all_level_forecasts = hierarchy.aggregate_ts(forecasts)

evaluate forecasting accuracy

# accuracy of reconciled forecasts
hierarchy.accuracy(test, forecasts, hist=train, measure=['mase', 'rmse'])

# accuracy of base forecasts
base_forecasts = model.generate_base_forecast(horizon=12)
hierarchy.accuracy_base(test, base_forecasts, hist=train, measure=['mase', 'rmse'])

because of the incoherence of base forecasts, base_forecasts are forecasts of all time series in the hierarchy, while coherent forecasts are forecasts of bottom time series.

Documentation

see documentation here https://angelpone.github.io/pyhts/.