FAIM Time-Series Forecasting icon

FAIM Time-Series Forecasting

Generate time-series forecasts using FAIM ML models (Chronos 2.0, FlowState, TiRex). Visit https://faim.it.com/ to learn more.
Join our community

Overview

This node generates time-series forecasts using FAIM machine learning models, specifically the Chronos 2.0 model. It is useful for predicting future values in a time series based on historical data, such as sales forecasting, demand planning, or resource allocation. Users provide input data as a JSON array, specify the forecast horizon (number of future time steps), and choose the output type (point forecasts or quantiles). The node then calls the FAIM API to obtain the forecast results.

Use Case Examples

  1. Forecasting sales for the next 24 hours based on past sales data.
  2. Predicting future demand for inventory management using quantile forecasts to understand uncertainty.

Properties

Name Meaning
Model The forecast model to use, fixed as Chronos 2.0 (hidden from user).
Input Data Time series data provided as a JSON array or an expression that resolves to a JSON array. This is the historical data used for forecasting.
Forecast Horizon Number of future time steps to forecast, must be between 1 and 1000.
Output Type Type of forecast output desired: either point forecasts or quantiles (probabilistic forecasts).
Quantiles JSON array of quantiles to forecast, applicable only when output type is quantiles and model is Chronos 2.0.

Output

JSON

  • json - The forecast results returned from the FAIM API, structured as JSON. This includes the forecasted values or quantiles depending on the output type.

Dependencies

  • FAIM API key credential required for authentication to the FAIM forecasting service.

Troubleshooting

  • Ensure the input data is a valid JSON array; invalid JSON will cause an error.
  • The FAIM API key must be provided and valid; missing or incorrect keys will cause authentication errors.
  • Quantiles must be a valid JSON array when using quantile output type; invalid format will cause errors.
  • Errors from the FAIM API are caught and presented with user-friendly messages.
  • Common error messages include invalid JSON input, missing API key, and API errors with explanations.

Links

  • FAIM API Documentation - Official documentation for the FAIM forecasting API used by this node.
  • FAIM Website - Homepage for FAIM, the provider of the forecasting models used in this node.

Discussion