Skip to content

paulozip/arauto

Repository files navigation

Documentation Status

Arauto

A interactive tool for time series experimentation and forecasting

Arauto is an open-source framework that aims to make it easier to model and experiment time series analysis and forecasting. Arauto offers a intuitive and interactive interface to explore different parameters for models using Autoregressive models (AR, ARMA, ARIMA, SARIMA, ARIMAX, and SARIMAX). More estimators and algorithms are on the way.

Blog post about Arauto

Arauto was featured on Towards Data Science's blog on Medium. Check it out.

Features

  • Support for exogenous regressors (independent variables)
  • Seasonal decompose that lets you know the Trend, Seasonality, and Resid of your data
  • Stationarity Test using Augmented Dickey-Fuller test
  • Customization of data transforming for stationarity: you can use from first difference to seasonal log to transform your data
  • ACF (Autocorrelation function) and PACF (Parcial correlation function) for terms estimation
  • Customize ARIMA terms or let Arauto choose the best for you based on your data
  • Grid search feature for parameters tuning
  • Code generation: at the end of the process, Arauto returns the code used to transform the data and train the model

Getting started

Arauto can be used in three different ways:

Web

If you are just curious about what you can do with Arauto, you can refer to this website. This version contains some toy datasets that you can use to check how Arauto works.

IMPORTANT: please be aware that this web version is a lightweight version of Arauto, with low computational power and with some features disable, since we are using a free-tier version of Heroku instance.. To understand the full potential of Arauto, use one of the installations version below.

Docker

Run the following commands to use Arauto with Docker (requires Docker and Docker-compose):

# Run the docker compose
docker-compose up --build

Local installation (requirements.txt)

Tip: we recommend you to use Anaconda environments

# Clone the repository
git clone https://github.com/paulozip/arauto.git
cd arauto

# If you're using Anaconda
conda create --name arauto_env
conda activate arauto_env

# Install dependencies
pip install -r requirements.txt

# Run Streamlit
streamlit run run.py

How to use

Please refer to our documentation to check the tutorial that we prepared to help you to build your own models.

Upload your data

A Upload file support will be added to Arauto, but you can use the Arauto REST API to send your dataset. Here's an example of how you can use it using cURL:

curl -X POST \
  http://SERVER_ADDRESS:5000/upload_file \
  -H 'content-type: multipart/form-data' \       
  -F file=@PATH_TO_YOUR_FILE

Example:

curl -X POST \
  http://0.0.0.0:5000/upload_file \
  -H 'content-type: multipart/form-data' \       
  -F file=@/home/my_user/Downloads/dataset.csv

Your data, your code

At the end of the process, Arauto will give you the source code used to make the exploration and training. Use this code to adapt your work and make modifications at will. PROTIP: Arauto is also a useful tool to generate boilerplate code for time series.

Next steps

Currently, this is an one-person project, but help is always welcome. You can suggest new ideas and features for us in our issue tracking. Requests will be prioritized by thumbs up emoji (👍).

If you want to collaborate with this project, all you need to do is fork this repository, make new additions and modifications, and open a PR. I will validate it ASAP.

Get updated

New features will be added to Arauto. If you want to know when a new version arrives, please, subscribe to Arauto’s newsletter to get informed about new things. We will NOT send spams or share your email. Only content related to Arauto will be sent.

A big thank you

This project would not be possible without Streamlit: an awesome tool to build ML tools. Please refer to their Github repository to know further, or you can also check their blog post