timesfm-forecasting

Name: timesfm-forecasting
Availability: InStock
Author: estrella-231

Generate zero-shot forecasts and quantile prediction intervals for univariate time series.

Updated May 24, 2026

Search Tags:#time series #forecasting #gpu #zero-shot #foundation model #quantile intervals #system preflight

Ready to run

npx skills add https://github.com/Estrella-231/Mathematical_modeling_tongmeng --skill timesfm-forecasting-estrella-231

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: timesfm-forecasting
Source: https://github.com/Estrella-231/Mathematical_modeling_tongmeng/tree/main/.agents/skills/timesfm-forecasting
Command: npx skills add https://github.com/Estrella-231/Mathematical_modeling_tongmeng --skill timesfm-forecasting-estrella-231

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires timesfm, torch, numpy, pandas, matplotlib, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

It solves the problem of producing accurate (and uncertainty-aware) forecasts for univariate time series without training a custom model, even when you do not know which classical method to tune.

Core Features & Use Cases

Zero-shot forecasting: Forecast any univariate numeric series using Google’s TimesFM foundation model without training.
Point forecasts + prediction intervals: Return a median-style point forecast plus calibrated quantile-based intervals (for example q10–q90 bands).
Practical safety guardrails: Includes a mandatory preflight system checker to validate RAM, GPU/VRAM, disk space, Python version, and installed packages before downloading/loading weights.

Quick Start

Use the timesfm-forecasting skill to forecast the next 24 steps of a single univariate sales series by running the mandatory preflight check, installing prerequisites, then compiling the model with a ForecastConfig and calling forecast() on your input array.

Frequently Asked Questions about timesfm-forecasting

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I generate zero-shot time series forecasts without training a model?▼

Zero-shot time series forecasting uses Google's TimesFM foundation model to predict univariate numeric series without training. You configure a ForecastConfig and call the forecast method on your input array to return point forecasts and calibrated quantile-based prediction intervals.

Can I get prediction intervals for my time series forecasts?▼

Yes, time series forecasting with TimesFM returns calibrated quantile-based prediction intervals, such as q10 to q90 bands, alongside a median-style point forecast. This provides probabilistic bounds for uncertainty-aware operational planning without requiring model training.

What are the system requirements for running TimesFM forecasting?▼

Running TimesFM forecasting requires passing a mandatory preflight system checker that validates available RAM, GPU or VRAM, disk space, Python version, and installed packages before downloading and loading the model weights to ensure operational safety.

Does zero-shot forecasting work for batch processing multiple time series?▼

Yes, zero-shot forecasting supports batch forecasting across many univariate time series simultaneously. You can process multiple sales, sensor, energy, demand, vitals, price, or weather series in a single workflow using TimesFM without training individual models.

How do I forecast the next 24 steps of a sales time series?▼

To forecast the next 24 steps of a sales series, run the mandatory preflight check, install prerequisites, compile the TimesFM model with a ForecastConfig, and call forecast on your input array to return 24 steps of point and quantile outputs.

What is the best way to forecast univariate data without tuning classical methods?▼

The best way to forecast univariate data without tuning classical methods is using a foundation model for zero-shot time series forecasting. TimesFM generates accurate predictions and uncertainty-aware intervals without requiring custom model training or method selection.