FILE: docs/index.md

DataKit Documentation

Welcome to DataKit, the Python utility package that eliminates repetitive data science boilerplate.

One-liner functions. Smart defaults. Full control when you need it.

What is DataKit?

Every data scientist rewrites the same code over and over:

• Checking missing values
• Cleaning column names
• Plotting distributions
• Running statistical tests
• Encoding categorical variables

DataKit wraps Pandas, NumPy, Matplotlib, Seaborn, and SciPy into simple one‑call functions. Beginners get professional results without knowing the internals. Experts can override every parameter.

Quick example

import datakit as dk
import pandas as pd

df = pd.read_csv('sales.csv')

# Full EDA in one line
dk.quick_eda(df)

# Clean it up
df = dk.clean_columns(df)
df = dk.fix_dtypes(df)
df = dk.auto_impute(df)

# Compare two groups
result = dk.compare_groups(
    df[df['region'] == 'North']['revenue'],
    df[df['region'] == 'South']['revenue']
)
print(result['interpretation'])

Features

Module	Key functions
pandas	auto_impute, fix_dtypes, clean_columns, profile_df, encode_categoricals
numpy	normalize, safe_divide, set_seed, assert_shape
matplotlib	quickplot, plot_grid, annotate_bars, ColorRegistry
seaborn	plot_corr, plot_distributions, set_theme
scipy	compare_groups, cohen_d, correct_pvalues, fit_and_plot
pipelines	quick_eda, groupby_chart

Installation

pip install datakit

Documentation structure

• Quickstart – Get started in 5 minutes
• API Reference – Detailed function documentation
• Configuration – Using .datakit.toml
• Examples – Jupyter notebooks

Design philosophy

DataKit follows six principles:

1. Sensible defaults with full override – Every function works with minimal arguments; advanced parameters are available as keyword arguments.
2. Transparent output – Functions tell you what they did (unless you set verbose=False).
3. Composability – Functions return standard objects (DataFrames, arrays, Axes) that chain naturally.
4. Fail loudly with helpful messages – Clear errors, not silent failures.
5. No side effects on input data – Operates on copies unless inplace=True.
6. Works in notebooks and scripts – Same code everywhere.

License

DataKit is released under the MIT license.