What data format does the selector accept?

The skill requires a CSV file with three specific columns: unique_id (identifier), ds (datetime timestamp), and y (numeric observations).

When does the skill choose StatsForecast over TimeGPT?

StatsForecast is selected for datasets with missing values, short data lengths (under 30 points), strong seasonal patterns, or very high series counts exceeding 100.

Can I set a custom forecast horizon?

Yes, while the default is 14 periods, you can specify a custom horizon using the --horizon flag when triggering the model selector script.

Is a Nixtla API key required for this skill?

An API key is only required if the skill determines that TimeGPT is the best model for your data. StatsForecast runs locally and does not require a key.

Does the skill handle data visualization?

Yes, by using the --visualize flag, the skill can generate a time_series_plot.png to help you inspect the input data and model fit.

Nixtla Time Series Model Selector

Name: Nixtla Time Series Model Selector
Author: intent-solutions-io

byintent-solutions-io

데이터 과학 및 ML

Automatically selects and executes the optimal forecasting engine between StatsForecast and TimeGPT based on your data's unique characteristics.

소개

The Nixtla Model Selector skill streamlines the time series forecasting workflow by intelligently choosing the most effective model for your specific dataset. It evaluates key metrics such as data length, seasonality, missing values, and series volume to decide between local statistical models like AutoARIMA and foundation models like TimeGPT. This eliminates manual experimentation and benchmarking, allowing developers to generate high-accuracy forecasts with a transparent decision rationale and standardized outputs for easy integration into data pipelines.

주요 기능

Heuristic-based selection between StatsForecast and TimeGPT
Automated detection of seasonality and data sparsity
Support for massive batch forecasting (100+ series)
0 GitHub stars
Standardized long-format CSV output for predictions
Detailed selection reporting with decision rationale

사용 사례

Benchmarking cloud-based AI models against classical statistical methods
Processing heterogeneous time series data with varying lengths and quality
Automating demand forecasting for inventory management systems

소개

주요 기능

Heuristic-based selection between StatsForecast and TimeGPT
Automated detection of seasonality and data sparsity
Support for massive batch forecasting (100+ series)
0 GitHub stars
Standardized long-format CSV output for predictions
Detailed selection reporting with decision rationale

사용 사례

Benchmarking cloud-based AI models against classical statistical methods
Processing heterogeneous time series data with varying lengths and quality
Automating demand forecasting for inventory management systems