LLM Driven AutoForecasting with Sktime’s `craft()`

Editor's take

Sktime, an open-source Python library for time series analysis, has introduced a new `craft()` function that leverages large language models (LLMs) to automate the process of forecasting time series data. This feature aims to simplify the selection and configuration of forecasting models by intelligently identifying patterns and suggesting appropriate algorithms.

The significance of this development lies in its potential to democratize advanced time series forecasting, making it accessible to users without deep statistical expertise. By abstracting away much of the model selection complexity, `craft()` could empower a wider range of data scientists and analysts to derive insights from temporal data, particularly in domains like finance, retail, and operations where accurate predictions are crucial.

Future developments to monitor include the performance benchmarks of `craft()` against human-expert model selection and traditional automated forecasting tools like Auto-ARIMA. The extent to which LLMs can generalize across diverse time series characteristics and the interpretability of the chosen models will be key indicators of its long-term impact.