Objective

The objective of the repository is to share generic functions that everybody can use on the day to day work.

Contribution Flow Summary

To contribute one should clone the repo, add the functions to share and then perform a pull request into the develop branch.

Pull requests must be well documented and fully working. A test example should be provided such that the code added can be tested on review.

Package Main Structure

ml_toolkit
|--clustering [extra]
|--db_interaction [extra]
|--deep_learning [extra]
|--feature_encoding [extra]
|--time_series [extra]
|--utils
|--webtools [extra]

Installation

There are two options:

Stable version:
pip install ml-toolkit --extra-index-url https://api.packagr.app/public

Directly from the repository
    Clone the repository, navigate to it and install

git clone https://github.com/TSPereira/ml_toolkit.git
cd ml_toolkit
pip install .

Please note that some extra options might need for you to define a find-links flag.
Options for installation as below.

pip install . (minimum installation)
pip install .[all] -f https://download.pytorch.org/whl/torch_stable.html (all extras)
pip install .[deep_learning] -f https://download.pytorch.org/whl/torch_stable.html
pip install .[<extra>] (specific extra) pip install .[<extra1>, <extra2>] (multiple extras)

To include in a requirements.txt you must add the flags before

numpy==1.18.5
--extra-index-url https://api.packagr.app/public -f https://download.pytorch.org/whl/torch_stable.html
ml-toolkit[all]==0.1.1

Note: Some internal packages might need additional dependencies (hdbscan needs either MVSC14.0++ or gcc for example)