An R package for tidy time series forecasting with models like ETS and ARIMA, integrated with the tidyverse.
fable is an R package for tidy time series forecasting, providing a collection of commonly used univariate and multivariate forecasting models like exponential smoothing (ETS) and automatic ARIMA. It solves the problem of integrating forecasting workflows with the tidyverse, offering tools to evaluate, visualize, and combine models in a consistent and reproducible manner.
Data analysts, data scientists, and researchers using R for time series analysis and forecasting, particularly those who work within the tidyverse ecosystem.
Developers choose fable for its seamless integration with the tidyverse, providing a consistent workflow for forecasting that includes model evaluation, visualization, and combination, along with support for both univariate and multivariate models.
Tidy time series forecasting
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Seamlessly works with tidyverse tools like dplyr and tsibble for data manipulation, as shown in the example code filtering and modeling time series data within a consistent workflow.
Provides automated ARIMA modelling, reducing the need for manual parameter specification and enabling quick forecasting without deep statistical expertise.
Includes both exponential smoothing (ETS) and ARIMA models for univariate and multivariate series, along with tools like SNAIVE for baseline comparisons.
Offers built-in functions like autoplot() for visualizing forecasts and model evaluation, integrating with ggplot2 for customizable plots without extra setup.
Installing fable requires a compiler, which can be problematic for users on systems without development tools or those unfamiliar with compiling R packages.
Primarily focuses on statistical methods like ETS and ARIMA, lacking support for newer techniques such as neural networks or gradient boosting, which might be needed for complex patterns.
Heavily relies on the tidyverse ecosystem (e.g., tsibble, dplyr), so users not committed to this framework may find it less flexible or face a steeper learning curve.