FAME Toolbox

Foundational AI Models for Time-Series Forecasting Select a model and dataset, configure parameters, then click Run.

Model

Dataset

Horizon

10 200

Compare All Models (Side-by-Side)

Interactive Plot (Plotly)

Enable for zoomable, hoverable multi-model comparison plots. Uses more memory.

Max Models

1 10

Window Size

Size of windows for non-overlapping or increasing window strategies.

40 500

Windowing Strategy

How to split time series into windows for training.

Target

The feature to predict.

Past Covariates

Historical-only features (future values unknown at prediction time).

Known Covariates

Features whose future values are known at prediction time (passed to the predictor as known_covariates).

Static Covariates

Per-item constants attached as static features. Models with native support use them directly; others get a covariate_regressor fallback.

Logs

Forecast

Current Plot Type

Optionally leave your email address to receive updates about the FAME Toolbox.

Status