This README is only about running the examples (notebooks + helper utilities). For Azure purchasing/deployment of the Managed App, see package-README.
New to inait Forecasting? See Why Choose inait Forecasting for a quick overview of models, ensembles, explainability, and industry use cases.
Table of Contents
Click the first Binder badge below to run directly on your browser. First launch of updated version of the code may take a few minutes while the image builds.
A) credentials.txt (default used by notebooks)
Edit the file credentials.txt at the root of the repo adding the following two lines, edit only the key value, replacing it with yours:
API_BASE_URL='https://api.forecasting.inait.ai/forecasting'
API_AUTH_KEY='your-api-key-string'- Notebooks: Set Credentials and Run ( 80 sec )
- Notebooks: Add new data and tailor the code ( 60 sec )
Notes:
- If it's the first time you run Jupyter notebooks, start by clicking on "Run" at the top menu and select "Run All Cells".
- If you get a
timeout error, please just restart the Kernel and give it another try.
Click the Codespaces badge. On first start, the dev container installs uv, runs make init to create .venv, and registers the Python (inait‑uv) kernel. Open notebook-examples/ and start any notebook.
A) credentials.txt (default used by notebooks)
Edit the file credentials.txt at the root of the repo adding the following two lines, edit only the key value, replacing it with yours:
API_BASE_URL='https://api.forecasting.inait.ai/forecasting'
API_AUTH_KEY='your-api-key-string'# 1) Install uv (Linux/macOS)
curl -LsSf https://astral.sh/uv/install.sh | sh
# 2) From the repo root, set up deps (creates .venv from pyproject.toml)
make init # or: uv sync
# 3) Launch JupyterLab
uv run jupyter labA) credentials.txt (default used by notebooks)
Create/edit a credentials.txt at the repo root:
API_BASE_URL='https://api.forecasting.inait.ai/forecasting'
API_AUTH_KEY='your-api-key'B) .env (supported via python‑dotenv)
Create .env at the repo root:
API_BASE_URL="https://api.forecasting.inait.ai/forecasting"
API_AUTH_KEY="your-api-key"Don’t have an endpoint yet? See package-README to deploy the Managed App and obtain credentials.
Run top‑to‑bottom:
| Notebook | What it shows |
|---|---|
notebook-examples/0_quickstart.ipynb |
Start here. Configure credentials, submit your first forecast using the airline passenger dataset, and visualize results. |
notebook-examples/1_advanced_model_evaluation.ipynb |
Compare inait-basic, inait-advanced, and inait-best models on the ETTh1 electricity transformer benchmark dataset. |
notebook-examples/2_energy_forecast_interpretability.ipynb |
Use inait explainability features to understand which factors drive energy price predictions. |
notebook-examples/3_sales_forecast_with_uncertainty.ipynb |
Sales forecasting with prediction intervals (uncertainty bands) using M5 competition data. |
Comprehensive forecasting workflows with backtesting and model comparison:
| Notebook | What it shows |
|---|---|
futurecomplete-examples/frontiers_articles.ipynb |
Scientific article submission forecasting using FMSA dataset with backtesting and model comparison. |
futurecomplete-examples/power.ipynb |
German electricity price forecasting with exogenous variables (load, consumption) and comprehensive evaluation. |
futurecomplete-examples/volatility_aapl.ipynb |
Financial volatility forecasting for Apple stock with advanced time series techniques. |
futurecomplete-examples/volatility_sp100.ipynb |
S&P 100 volatility forecasting demonstrating portfolio-level risk modeling. |
Small CSVs live in data/:
data/airline.csv– Classic monthly airline passengers time series (Box & Jenkins dataset)data/etth1.csv/data/etth1_small.csv– ETTh1 energy transformer dataset (full and small versions)data/M5_store_CA_1.csv– M5 competition dataset (single-store sales sample)data/power_day_ahead.csv– German day-ahead hourly electricity prices with exogenous factors
data/dataset_GKYZ_2016.csv– Financial dataset for GKYZ volatility modelling with 100 stocksdata/dataset_GKYZ_2016_AAPL.csv– Apple stock data for volatility modeling
data/FMSA_Articles.csv– Frontiers in Marine Science article submissions (academic forecasting)
Expected format (simplified):
- A timestamp column with consistent frequency (hourly, daily, …)
- One or multiple numeric columns: one or more target columns to forecast and optional exogenous variables all aligned to the same timestamps
All examples use the inait-forecasting-client package which provides:
from inait_forecasting_client import (
predict, # Submit forecasting jobs
plot, # Visualize results
read_file, # Load data with proper formatting
backtest, # Model evaluation and validation
compare, # Compare multiple models
)Key Features:
- Simple API: Submit forecasts with just a few lines of code
- Auto-visualization: Built-in plotting for results and diagnostics
- Flexible data loading: Supports CSV, Excel, and pandas DataFrames
- Model comparison: Built-in backtesting and performance evaluation (scores included in results)
- Kernel mismatch (Codespaces): ensure the notebook kernel is Python (inait‑uv).
- Import errors in terminal:
uv sync && source .venv/bin/activate. - Auth errors (401/403): check
API_AUTH_KEYand tenant for your endpoint. - Background jobs: some examples poll until completion—keep the cell running.
- Issues: https://github.com/inait-external/inait-forecast-docs/issues
- Email: contact@inait.ai
Next: Azure purchase & deployment → package-README