Automated Machine Learning Benchmark

Table of Contents

• Background
• Install
• Usage
• API
• Contributing
• License

Background

This repo is a suite of automated machine learn(autoML) frameworks in python. It provides benchmarking with open source autoML frameworks. Caution: due to some of the package compatiablity, this suite is only tested in linux system. It may work with windows linux subsystem, but not verified.

Any optional sections

Install

• Install anaconda or miniconda [here]:(https://docs.conda.io/projects/conda/en/latest/user-guide/install/index.html)

• Copy/clone all content in this repo

• Set up environment with yml file as in [here]: (https://docs.conda.io/projects/conda/en/latest/user-guide/tasks/manage-environments.html#creating-an-environment-from-an-environment-yml-file)

$ cd framework/
$ conda env create -f autosklearn.yml

• Activate python environment with conda:

$ conda activate autosklearn

• Check you data directectory is as defined in src/automl/autobench.py files
• test run

$ cd automl
$ bash run.sh

Usage

Input and output files

Examples of run open source automated machine learning frameworks

Baseline model building with automated machine learning frameworks

Automated benchmarking with selected frameoworks/datasets

Note: The license badge image link at the top of this file should be updated with the correct :user and :repo.

Any optional sections

API

Any optional sections

More optional sections

Contributing

See the contributing file!

Any optional sections