fable

The R package fable provides methods and tools for displaying and analysing univariate time series forecasts including exponential smoothing via state space models and automatic ARIMA modelling. Data, model and forecast objects are all stored in a tidy format.

Installation

You can install the development version from GitHub

# install.packages("devtools")
devtools::install_github("tidyverts/fable")

Example

library(fable)
library(tsibbledata)
UKLungDeaths %>%
  model(ets = ETS(log(mdeaths))) %>%
  forecast
#> # A fable: 24 x 5 [1M]
#> # Key:     .model [1]
#>    .model    index    .h mdeaths .distribution    
#>    <chr>     <mth> <dbl>   <dbl> <dist>           
#>  1 ets    1980 Jan     1   1832. t(N(7.5, 0.0095))
#>  2 ets    1980 Feb     2   1854. t(N(7.5, 0.0095))
#>  3 ets    1980 Mar     3   1732. t(N(7.5, 0.0094))
#>  4 ets    1980 Apr     4   1444. t(N(7.3, 0.0089))
#>  5 ets    1980 May     5   1155. t(N(7.0, 0.0084))
#>  6 ets    1980 Jun     6   1050. t(N(7.0, 0.0082))
#>  7 ets    1980 Jul     7   1000. t(N(6.9, 0.0080))
#>  8 ets    1980 Aug     8    915. t(N(6.8, 0.0078))
#>  9 ets    1980 Sep     9    915. t(N(6.8, 0.0078))
#> 10 ets    1980 Oct    10   1081. t(N(7.0, 0.0082))
#> # … with 14 more rows

You can read more about the functionality of this package and the ideas behind it here: https://tidyverts.github.io/tidy-forecasting-principles/