Skip to content

Quickstart

Overview

Bowline is split up into Preprocessors and PostProcessors and are stored in bowline.preprocessors and bowline.postprocessors, respectively. They're all made with a simple implementation as the default but contain extensive advanced configurations for processing. Lets start with the simplest one.

Load in processor

Lets load in the StandardPreprocessor.

from bowline.preprocessors import StandardPreprocessor

Process Data

Now we read in a csv, pass it to StandardPreprocessor with some dataset specific information, and finally process the dataset for a given target variable.

import pandas as pd

raw_data = pd.read_csv('path/to/your/file')
preprocessor = StandardPrepreocessor(
    data = data,
    numerical_features = ["age", "capital-gain"],
    binary_features = ["sex"],
    categoric_features = ["occupation", "education"]
)
processed_data = preprocessor.process(target="sex")

Further configuration

All Preprocessors and Postprocessors have additional parameters to configure how the data is processed. Currently, there is no official documentation for it but each class has extensive docstings to help with implementation.