dsbro 0.1.0

Your Data Science Bro. One import away.

dsbro

Your Data Science Bro. One import away.

dsbro is a lightweight Python toolkit for notebook-heavy data science work. It is built for the repeated workflow most people have in Kaggle, Colab, and local Jupyter notebooks: setup, load data, inspect it, clean it, visualize it, and train a baseline model fast.

What dsbro covers

• dsbro.utils: notebook setup, seeding, timers, system info, downloads, simple parallel work
• dsbro.io: file loading, saving, previews, directory trees, file search, submission helpers
• dsbro.eda: overview tables, missing-value analysis, drift checks, target analysis, comparisons
• dsbro.prep: encoding, scaling, missing-value filling, feature engineering, memory reduction
• dsbro.viz: themed matplotlib and optional plotly charts for fast notebook visuals
• dsbro.metrics: classification and regression metrics in one place
• dsbro.ml: model comparison, cross-validation, training, tuning, stacking, pseudo-labeling
• dsbro.text: text cleaning, tokenization, word frequencies, and TF-IDF features

Current status

Implemented now:

• Core package scaffold and packaging
• utils, io, eda, prep, viz, metrics, ml, and text
• Built-in help and about/version entry points
• Tests across the implemented modules
• Quickstart notebook in examples/quickstart.ipynb

Still planned:

• Final polish for docs/examples
• Additional ML/deep-learning extras over time

Installation

From PyPI:

pip install dsbro

From GitHub:

pip install git+https://github.com/muhammadibrahim313/dsbro.git

For local development:

pip install -e ".[dev]"

Optional extras:

pip install -e ".[ml]"
pip install -e ".[plotly]"
pip install -e ".[all]"

PyPI packaging is scaffolded, but this repository is still in active buildout.

Quick example

import dsbro
import pandas as pd

dsbro.setup()

train = pd.DataFrame(
    {
        "age": [22, 35, 41, 28],
        "city": ["lahore", "karachi", "lahore", "islamabad"],
        "purchased": [0, 1, 1, 0],
    }
)

overview = dsbro.eda.overview(train)
processed, report = dsbro.prep.auto_preprocess(train, target="purchased")
leaderboard = dsbro.ml.compare(train, target="purchased", cv=2)

Help system

dsbro includes a built-in cheatsheet:

dsbro.help()
dsbro.help("viz")
dsbro.help("encode")
dsbro.about()
dsbro.version()

Notebook example

The repository includes a walkthrough notebook:

• examples/quickstart.ipynb

It demonstrates:

• dsbro.setup()
• dsbro.eda.overview()
• dsbro.prep.datetime_features()
• dsbro.prep.text_features()
• dsbro.prep.auto_preprocess()
• dsbro.viz.bar()
• dsbro.ml.compare()

Development

pytest tests/ -v
ruff check dsbro/ tests/
ruff format dsbro/ tests/
python -m build

Roadmap

• Expand example notebooks
• Add GitHub Actions CI
• Publish to TestPyPI, then PyPI
• Continue polishing module docs and tutorial coverage

License

MIT. See LICENSE.