seasonal naive confidence intervals are incorrect

[andrewscottm]

What happened + What you expected to happen

The seasonal naive confidence interval for the last time in the forecasted horizon is used for all times.

statsforecast/python/statsforecast/models.py

Line 3305 in 12f6654

k = np.floor((h - 1) / self.season_length)

Should be k = np.floor(np.arange(h) / self.season_length) as in the R version.

When the time horizon increases, the first forecasted time's confidence interval increases. It should be constant.

Versions / Dependencies

statsforecast 1.7.6
pandas 2.2.2
Python 3.11.9 (main, Apr 19 2024, 11:44:45) [Clang 14.0.6 ]
MacOS 14.6.1

Reproducible example

import pandas as pd
from statsforecast import StatsForecast
from statsforecast.models import SeasonalNaive

df = pd.DataFrame({
    'unique_id': '0',
    'ds': pd.date_range(start='2000-01-01', periods=10, freq='1D'),
    'y': [1., 4., 5., 5., 1., 4., 5., 5., 7., 3.]
})

m = StatsForecast(models=[SeasonalNaive(season_length=7)], freq='1D', n_jobs=1, verbose=True)
m.fit(df=df)
print(m.predict(h=1, level=[95]).iloc[0, :])
print(m.predict(h=8, level=[95]).iloc[0, :])
print(m.predict(h=15, level=[95]).iloc[0, :])

unique_id 0
ds 2000-01-11 00:00:00
SeasonalNaive 5.0
SeasonalNaive-lo-95 -1.093776
SeasonalNaive-hi-95 11.093776

unique_id 0
ds 2000-01-11 00:00:00
SeasonalNaive 5.0
SeasonalNaive-lo-95 -3.6179
SeasonalNaive-hi-95 13.6179

unique_id 0
ds 2000-01-11 00:00:00
SeasonalNaive 5.0
SeasonalNaive-lo-95 -5.554729
SeasonalNaive-hi-95 15.554729

Issue Severity

Medium: It is a significant difficulty but I can work around it.