potpourri 0.14.0

model zoo of different preconfigured algorithms

potpourri

Table of Contents

• Project Requirements
• Folder Structure
• Installation
• Usage
• Commands
• Support
• Contributing

Project Requirements

• apply different algorithms to a dataset as a batch script
• store evalutations (results, run times) in a database

Folder Structure

• potpourri -- different model implementations as python module. Each module contains three objects:
  • model -- a sklearn Pipeline to fit and predict
  • hyper -- dictionary with hyperparameters for sklearn's RandomizedSearchCV,
  • meta -- a python dict with further information
• verto -- Feature Engineering. Each module contain two objects
  • trans -- a sklearn pipeline to transform data
  • meta -- a python dict with further information
• seasalt -- contains different utility, glue, etc. functions and classes
• nbs -- notebooks to try, check, profile, etc. each model
• datasets -- demo datasets

Installation

The potpourri git repo is available as PyPi package

pip install potpourri

Usage

Check the nbs folder for notebooks.

Commands

• Check syntax: flake8 --ignore=F401,E251
• Remove .pyc files: find . -type f -name "*.pyc" | xargs rm
• Remove __pycache__ folders: find . -type d -name "__pycache__" | xargs rm -rf
• Remove Jupyter checkpoints: find . -type d -name ".ipynb_checkpoints" | xargs rm -rf
• Upload to PyPi with twine: python setup.py sdist && twine upload -r pypi dist/*

Othe helpful commands

• Find package folders: python -c 'from setuptools import find_packages; print(find_packages())'

Support

Please open an issue for support.

Contributing

Please contribute using Github Flow. Create a branch, add commits, and open a pull request.