parthivnaresh / facilyst

Facilyst is a library that makes using data science and machine learning tools easier.

Facilyst

Facil • yst

Facilis - Latin for easy, ready, or quick.

Analyst - Leaving this blank because I don't think anyone knows.

As a software developer in machine learning, I often find myself in the awkward position of building features and expanding use cases without easy access to data to test them with. Sometimes it could be as simple as a Pandas DataFrame with different data types just so I have something basic to throw at my code, and in other cases I might need a sinusoidal wave with varying frequency and amplitude consisting of an upwards trend projected over a staggered datetime horizon.

Instead of having to constantly look up the right functions in order to meet these data needs, and then search through StackOverflow or documentation to figure out how to plot it in a visually appealing way, I wanted a library that could provide a simple interface that made it easy to facilitate the work of an analyst/developer.

Installation

Install from PyPI

pip install facilyst

Version - 0.0.2

This is a work in progress and there are many features to be added.

facilyst's Issues

Add optimizers for models

Bayesian, Hyperopt, Optuna, etc

Fix path to read dataset in `get_dataset` when running `make test`

Running make test causes

dataset = pd.read_csv(
        f"{os.path.dirname(os.getcwd())}/datasets/{type_}/{dataset_name}.csv"
    )

to refer to a wrong path preventing these tests from being run locally from the terminal.

Add ReadTheDocs for documentation

Add functionality for aggregating datetime columns

Convert Year, Month, Day, etc columns into a date time column.

Add update dependency bot to GitHub Actions

Update readme for new installation instructions

Add git lfs to readme