Extends a fitted FASSTER model by filtering new observations through the existing state space model. The model's states and parameters are updated sequentially as new data arrives, allowing for online learning without refitting from scratch.
Usage
# S3 method for class 'FASSTER'
stream(object, new_data, specials = NULL, ...)Value
An updated FASSTER model object with:
Extended state estimates incorporating the new data
Updated model variance
Appended fitted values and residuals
Updated DLM components for future forecasting
Details
The streaming process:
Constructs the design matrix from new data
Applies the Kalman filter to sequentially update states
Updates model variance based on all residuals
Prepares the model for subsequent forecasting or streaming
Examples
library(tsibble)
library(fasster)
# Fit initial model on training data
fit <- as_tsibble(head(USAccDeaths, -12)) |>
model(fasster = FASSTER(value ~ trend() + season("year")))
tidy(fit)
#> # A tibble: 13 × 4
#> .model term estimate std.error
#> <chr> <chr> <dbl> <dbl>
#> 1 fasster "season(\"year\")" -167. 1.91
#> 2 fasster "season(\"year\")" -215. 1.92
#> 3 fasster "season(\"year\")" 278. 1.92
#> 4 fasster "season(\"year\")" -64.0 1.91
#> 5 fasster "season(\"year\")" 991. 1.92
#> 6 fasster "season(\"year\")" 1515. 1.92
#> 7 fasster "season(\"year\")" 988. 1.92
#> 8 fasster "season(\"year\")" 264. 1.92
#> 9 fasster "season(\"year\")" -488. 1.91
#> 10 fasster "season(\"year\")" -695. 1.91
#> 11 fasster "season(\"year\")" -1593. 1.93
#> 12 fasster "trend()" 9830. 2.92
#> 13 fasster "trend()" -59.7 1.52
tail(fitted(fit), 20)
#> # A tsibble: 20 x 3 [1M]
#> # Key: .model [1]
#> .model index .fitted
#> <chr> <mth> <dbl>
#> 1 fasster 1976 May 9187.
#> 2 fasster 1976 Jun 9240.
#> 3 fasster 1976 Jul 9513.
#> 4 fasster 1976 Aug 9305.
#> 5 fasster 1976 Sep 8052.
#> 6 fasster 1976 Oct 8263.
#> 7 fasster 1976 Nov 7973.
#> 8 fasster 1976 Dec 7769.
#> 9 fasster 1977 Jan 7774.
#> 10 fasster 1977 Feb 7483.
#> 11 fasster 1977 Mar 7870.
#> 12 fasster 1977 Apr 7974.
#> 13 fasster 1977 May 9014.
#> 14 fasster 1977 Jun 9382.
#> 15 fasster 1977 Jul 10268.
#> 16 fasster 1977 Aug 9798.
#> 17 fasster 1977 Sep 8430.
#> 18 fasster 1977 Oct 8651.
#> 19 fasster 1977 Nov 8109.
#> 20 fasster 1977 Dec 8541.
# Stream new data through the model
fit_updated <- fit |>
stream(as_tsibble(tail(USAccDeaths, 12)))
tidy(fit_updated)
#> # A tibble: 13 × 4
#> .model term estimate std.error
#> <chr> <chr> <dbl> <dbl>
#> 1 fasster "season(\"year\")" -167. 1.91
#> 2 fasster "season(\"year\")" -215. 1.92
#> 3 fasster "season(\"year\")" 278. 1.92
#> 4 fasster "season(\"year\")" -64.0 1.91
#> 5 fasster "season(\"year\")" 991. 1.92
#> 6 fasster "season(\"year\")" 1515. 1.92
#> 7 fasster "season(\"year\")" 988. 1.92
#> 8 fasster "season(\"year\")" 264. 1.92
#> 9 fasster "season(\"year\")" -488. 1.91
#> 10 fasster "season(\"year\")" -695. 1.91
#> 11 fasster "season(\"year\")" -1593. 1.93
#> 12 fasster "trend()" 9830. 2.92
#> 13 fasster "trend()" -59.7 1.52
tail(fitted(fit_updated), 20)
#> # A tsibble: 20 x 3 [1M]
#> # Key: .model [1]
#> .model index .fitted
#> <chr> <mth> <dbl>
#> 1 fasster 1977 May 9014.
#> 2 fasster 1977 Jun 9382.
#> 3 fasster 1977 Jul 10268.
#> 4 fasster 1977 Aug 9798.
#> 5 fasster 1977 Sep 8430.
#> 6 fasster 1977 Oct 8651.
#> 7 fasster 1977 Nov 8109.
#> 8 fasster 1977 Dec 8541.
#> 9 fasster 1978 Jan 7963.
#> 10 fasster 1978 Feb 7300.
#> 11 fasster 1978 Mar 7756.
#> 12 fasster 1978 Apr 8058.
#> 13 fasster 1978 May 8996.
#> 14 fasster 1978 Jun 9494.
#> 15 fasster 1978 Jul 10666.
#> 16 fasster 1978 Aug 9504.
#> 17 fasster 1978 Sep 8721.
#> 18 fasster 1978 Oct 9493.
#> 19 fasster 1978 Nov 8656.
#> 20 fasster 1978 Dec 9035.