compute_multi_window_metrics: Score a forecast over several evaluation windows at once

In forecastsuite: Multi-Model Time Series Forecasting Toolkit and Local App

Score a forecast over several evaluation windows at once

Description

A single aggregate score hides where a model actually fails – it can look strong on the full training range while drifting badly over the most recent months. Scores Train / Last 6 Months / Last 2 Years / Test (held out) using safe_compute_metrics(), skipping windows the series is too short to contain rather than reporting them as NA.

Usage

compute_multi_window_metrics(forecast_df, train_df, test_df, date_agg = "day")

Arguments

forecast_df	A tibble with ds/yhat columns.
train_df	A tibble with ds/y columns – the training data.
test_df	A tibble with ds/y columns – the held-out test data.
date_agg	Aggregation frequency (currently unused directly, kept for interface consistency with the rest of the package).

Value

A tibble with columns Set, Metric, Value – one block per window that applied.

Examples

ds <- as.Date("2024-01-01") + 0:29
train_df <- tibble::tibble(ds = ds[1:20], y = 1:20)
test_df <- tibble::tibble(ds = ds[21:30], y = 21:30)
forecast_df <- tibble::tibble(ds = ds, yhat = 1:30 + 0.5)
compute_multi_window_metrics(forecast_df, train_df, test_df)

forecastsuite documentation built on Oct. 18, 2026, 5:08 p.m.