Quick start¶
Install skforecast-ai and verify the setup works. For a step-by-step walkthrough of your first real forecast, continue to Your first forecast.
New to forecasting with machine learning? skforecast's Introduction to forecasting covers the fundamentals this documentation assumes.
Install¶
Core library — no API key needed, runs entirely offline:
pip install skforecast-ai
Adds the optional LLM reasoning layer for explanations and Q&A:
pip install "skforecast-ai[llm]"
For development or contributing:
git clone https://github.com/skforecast/skforecast-ai.git
cd skforecast-ai
pip install -e ".[dev]"
Smoke test¶
Run the snippet below. If it prints a predictions table and a metrics row, the installation is working.
import pandas as pd
from skforecast_ai import ForecastingAssistant
from skforecast.datasets import load_demo_dataset
data = load_demo_dataset(verbose=False)
assistant = ForecastingAssistant(llm=None)
result = assistant.forecast(data=data, target="y", steps=12, test_size=12)
print(result.predictions) # forecast for the held-out test window
print(result.metrics) # evaluation metrics: MAE, MSE, MASE
print(result.code) # the skforecast script that produced this result
Runs locally by default
The smoke test runs in deterministic mode: no LLM, no network access, and no configuration required.
Next steps¶
- Your first forecast: the same call, explained step by step.
- Agentic forecasting: End-to-end walkthrough on a real hourly dataset.