Do I need an API key to use TimeGPT?

Yes, while statistical models like AutoARIMA and SeasonalNaive work locally, you must provide an NIXTLA_API_KEY to access TimeGPT features.

How does the skill handle data preparation?

It enforces the Nixtla standard schema: 'unique_id' for series identification, 'ds' for timestamps, and 'y' for the target values, ensuring compatibility across all models.

What Python libraries are required for this skill?

You will need Python 3.8+ and should install statsforecast, mlforecast, nixtla, and utilsforecast to utilize the full range of capabilities.

Can I compare multiple forecasting models at once?

Absolutely. The skill is designed to run benchmarks that compare classical statistical baselines, ML-based models, and LLM-based forecasting simultaneously.

Nixtla Time Series Lab

Name: Nixtla Time Series Lab
Author: intent-solutions-io

byintent-solutions-io

データサイエンスとML

Generates production-grade time series forecasts, model benchmarks, and predictive pipelines using the Nixtla ecosystem.

概要

The Nixtla TimeGPT Lab skill transforms Claude into a specialized forecasting expert, automating the implementation of StatsForecast, MLForecast, and TimeGPT. It streamlines the end-to-end forecasting workflow—including data schema mapping, model selection, and performance evaluation—to ensure high-accuracy predictions. Ideal for demand planning and predictive analytics, this skill generates standardized, production-ready Python code and visual comparisons to help developers deploy state-of-the-art forecasting models with minimal manual effort.

主な機能

Production-ready Python code generation and FastAPI service scaffolding
0 GitHub stars
Comprehensive model benchmarking with sMAPE, MASE, and MAE metrics
Standardized data schema mapping (unique_id, ds, y) for consistency
Integrated visualization tools for forecast comparison and confidence intervals
Automated forecasting using StatsForecast, MLForecast, and TimeGPT

ユースケース

Demand planning and inventory optimization for retail or logistics
Benchmarking classical statistical models against AI foundation models
Rapidly prototyping and deploying scalable time series APIs

概要

主な機能

Production-ready Python code generation and FastAPI service scaffolding
0 GitHub stars
Comprehensive model benchmarking with sMAPE, MASE, and MAE metrics
Standardized data schema mapping (unique_id, ds, y) for consistency
Integrated visualization tools for forecast comparison and confidence intervals
Automated forecasting using StatsForecast, MLForecast, and TimeGPT

ユースケース

Demand planning and inventory optimization for retail or logistics
Benchmarking classical statistical models against AI foundation models
Rapidly prototyping and deploying scalable time series APIs