tsibblefable

Tsibblefable is a term used to describe a unified approach to time series forecasting that combines the time-aware data structure of the tsibble package with the forecasting modelling capabilities of the fable package in R. The concept seeks to provide a coherent, tidyverse–friendly workflow for preparing, modeling, and forecasting time series data.

Core ideas include representing data with tsibble objects that capture temporal indices and grouping keys, while applying fable–style model specifications to produce forecasts. A tsibblefable workflow supports group-wise modelling, handling missing values, and comparing forecast accuracy across models. It also anticipates features such as hierarchical forecasting and flexible horizon definitions.

Relationship to packages: tsibble supplies the data container and indexing semantics; fable supplies models, forecasts, and diagnostics. The tsibblefable concept envisions a bridge in which these components interoperate seamlessly, enabling a single pipeline that runs models, produces forecasts, and provides plotting and reporting through the same interfaces.

Implementation status: tsibblefable is not an official standalone package. It appears in documentation and tutorials as a description of best practices for using tsibble and fable together. Some contributors discuss a formal extension or wrapper to expose a unified API, but there is no widely adopted standard as of now.

Impact: The idea is well aligned with tidy data principles and reproducible forecasting workflows, informing how users structure data and compare model performance across time series groups.