A starter pack for data science projects.
This repo is designed to bootstrap data science projects by providing a working environment for the development of Jupyter Notebooks. Useful classes from past projects are collected in src/datascience_starter. Classes are made available through the starter_pack package. Please use these classes to aid the development of jupyter notebooks within the notebooks dir.
Jupyter Notebooks are the defacto tool for exploring data and presenting work to peers and technical stakeholders. Unforntuately, notebooks are not the best way to re-use code between projects. This is what the datascience-starter repo helps structure. Code snippets for gridsearch, model diagnosis and data processing (etc.) are not often critical to the notebooks presentation. However, they are often a core step in the analysis pipeline. By abstracting these code snippits to reusable classes, one can develop a reusable codebase and cleaner notebooks.
To install the datascience-starter package,
git clone https://github.com/jordanparker6/datascience-startercd datascience-starterpip install .ormake install
The datascience-starter package can be used in any python file or jupyter notebook once installed using import datascience_starter. To take advantage of the project structure, all project working should be completed in the notebooks directory and notebooks should harness the reusable code maintained in the datascience-starter package.
Other usage comands are as follows.
- To build the documentation for the datascience-starter package
make docs. - To run tests over the datasceince-starter package
make tests. - To run linting over the datascience-starter package
make lint.