Forecast univariate time series with TimesFM using zero-shot inference.
npx skills add https://github.com/aleph23/Natasha --skill timesfm-forecasting-aleph23
Please help me install this Agent Skill. Skill: timesfm-forecasting Source: https://github.com/aleph23/Natasha/tree/main/skills/timesfm-forecasting Command: npx skills add https://github.com/aleph23/Natasha --skill timesfm-forecasting-aleph23
TimesFM enables zero-shot forecasting of a single time series without training, delivering fast probabilistic forecasts and safe resource checks.
Run the preflight check, load TimesFM, compile a ForecastConfig, and forecast a univariate time series.
High-intent search queries and answers about installing and using this skill.
Zero-shot time-series forecasting uses pre-trained models like TimesFM to generate future predictions instantly without requiring historical training data. This Skill applies TimesFM to produce probabilistic forecasts with calibrated prediction intervals directly from your existing inputs.
Yes, you can forecast time-series data from a pandas DataFrame or numpy array. This Skill accepts CSV files, DataFrames, and numpy arrays as inputs, returning univariate point forecasts along with calibrated prediction intervals for your datasets.
Yes, TimesFM supports covariate workflows for time-series forecasting through a dedicated forecast_with_covariates process. This allows you to incorporate external influencing variables alongside your primary univariate series to refine prediction accuracy.
To prevent RAM or GPU crashes when loading a large forecasting model, this Skill runs a mandatory preflight system check. It evaluates available memory and GPU resources before initializing TimesFM to ensure safe execution.
The best way to forecast sales or sensor data without historical training is using zero-shot inference. TimesFM analyzes univariate time series from CSV or DataFrame inputs to deliver fast probabilistic predictions suitable for these use cases.
SciMate-AIForecast univariate time series with TimesFM and quantile intervals.
jasrajtulsiForecast univariate time series with TimesFM zero-shot inference.
dralkhForecast univariate time series with TimesFM and generate prediction intervals.
nahisahoFine-tune pre-trained models on domain-specific scientific datasets with transfer learning workflows.
