AM_filler

Automatic Missing Value Filler - Fill missing values in datasets intelligently with one line of code.

🚀 Overview

AM_filler is a Python library that automatically detects column types and fills missing values using the best strategy — without any user configuration. It saves time and prepares datasets for ML/DL tasks in one line of code.

Why AM_filler?

📦 Installation

# Clone the repository
git clone https://github.com/yourusername/AM_filler.git
cd AM_filler

# Install in development mode
pip install -e .

⚡ Quick Start

from am_filler import AMFiller
import pandas as pd

# Your DataFrame with missing values
df = pd.read_csv("your_data.csv")

# One line to fill all missing values!
df_clean = AMFiller().fit_transform(df)

That's it! AM_filler automatically:

• Detects column types (numeric, categorical, text)
• Chooses the best imputation strategy
• Fills all missing values
• Logs what was done

🧠 How It Works - The Logic

AM_Filler uses intelligent algorithms to determine how to fill missing data:

1. Numeric Data (Intelligence)

It checks the distribution of data before deciding:

• Normal Distribution: Uses MEAN.
• Skewed / Outliers: Uses MEDIAN.

(See the graph above for visual representation)

2. Context-Aware Text Filling

Unlike other libraries that drop text or fill with "Missing", AM_Filler uses context-aware templates:

• Description: "Information not available."
• Feedback: "No review provided."
• etc.

3. Categorical Data

• Uses MODE (Most frequent value).

📁 Project Structure

AM_filler/
├── am_filler/
│   ├── __init__.py      # Public API exports
│   ├── core.py          # Main AMFiller class
│   ├── numeric.py       # Numeric imputation (smart mean/median)
│   ├── categorical.py   # Categorical imputation (mode)
│   ├── text.py          # Text imputation (sentences)
│   └── utils.py         # Helper functions
├── examples/
│   └── test_notebook.ipynb # Interactive Demo!
├── docs/
│   └── images/          # Documentation assets
├── README.md
├── pyproject.toml
└── setup.py

🧪 Running Tests

# Install test dependencies
pip install pytest

# Run all tests
pytest tests/ -v

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.