Bluemist AI

Bluemist AI is a low code machine learning library written in Python to develop, evaluate and deploy automated Machine Bluemist AI is an open source, low code machine learning library written in Python to develop, evaluate and deploy automated Machine Learning models and pipleines.

It acts as a wrapper service on top of sklearn, numpy, pandas, mlflow and FastAPI. Visualization are created using pandas-profiling, sweetviz, dtale and autoviz.

Features

• Native integration for data extraction with MySQL, PostgreSQL, MS SQL, Oracle, MariaDB, Amazon Aurora and Amazon S3
• Exploratory Data Analysis (EDA)
• Data preprocessing
• Trains data across multiple algorithms and provide comparison metrics
• Hyperparameter tuning
• Experiment tracking
• API deployment

For the detailed list of supported and upcoming features, visit https://www.bluemist-ai.one

Full documentation is available @ https://bluemist-ai.readthedocs.io

User installation

Method 1

To install minimal version of the package with hard dependencies listed in requirements.txt

pip install -U bluemist

To install the complete package including optional dependencies listed in requirements-optional.txt. Refer Minimal package vs Full package for more details.

pip install -U bluemist[complete]

Method 2 (recommended)

It is advised to setup a separate python environment to avoid conflicts with package dependencies. This can be done as follows :

• Install the package virtualenv

pip install virtualenv

• Create a separate directory where bluemist environment will be created

mkdir /path/to/bluemist-ai
cd /path/to/bluemist-ai

• Create the bluemist environment

virtualenv bluemist-env

• Activate the environment and install bluemist

source bluemist-env/bin/activate
pip install -U bluemist

Method 3

bluemist package can be installed using pipx utility. It automatically creates an isolated environment to run the bluemist package

pip install pipx
pipx install bluemist
pipx upgrade bluemist

Minimal package vs Full package

Below functionalities are available only with complete package installation

• Data extraction from RDBMS or cloud
• EDA Visualizations using dtale and sweetviz

Alternatively a single optional package can be installed using the pip command. For example, if you would like to extract data from Amazon S3 but do not wish to install other optional packages :

pip install boto3

License

Bluemist AI source code is licensed under the MIT License

See Third Party Libraries for license details of third party libraries included in the distribution.