keeya

Open-source Python library that converts natural language into executable code for data analysis, cleaning, visualization, and machine learning. Powered by Google Gemini. All generated code is returned for review before execution.

Install

pip install keeya

Setup

import keeya
keeya.setup()  # enter your Gemini API key (free at aistudio.google.com/app/apikey)

Usage

import pandas as pd
import keeya

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

# generate code from natural language
code = keeya.generate("create a function to calculate fibonacci numbers")

# clean your data — handles missing values, duplicates, type issues
cleaned_code = keeya.clean(df)

# get summary statistics and insights
analysis = keeya.analyze(df)

# generate visualizations based on your data
viz_code = keeya.visualize(df)

# build a complete ML pipeline
ml_code = keeya.train(df, target='target_column')

Every function returns code for you to review and execute. Context-aware — generation is based on your actual dataframe schema, not generic templates.

Run keeya.help() for detailed usage and examples.

License

MIT