AutoDataMind 🧠

Zero-Code Automated Data Science

AutoDataMind is your Native Intelligence Layer for data science. No Pandas knowledge required. No ML expertise needed. Just simple function calls for complete automation.

🌍 Philosophy

Data science should be accessible to everyone, everywhere.

AutoDataMind democratizes data science for:

• Emerging Markets: Africa, Asia, Latin America
• Small Businesses: No data science budget
• Students: Learning without complexity
• Non-Technical Users: Business analysts, researchers

✨ Features

🎯 Zero-Code Automation

import autodatamind as adm

# Automatic data analysis - ONE line
adm.analyze("sales.csv")

# Automatic ML training - ONE line
model = adm.autotrain("sales.csv", target="revenue")

# Automatic dashboard - ONE line
adm.dashboard("sales.csv")

# Automatic deep learning - ONE line
dl_model = adm.auto_deep("data.csv", target="category")

# Automatic insights report - ONE line
report = adm.generate_insights("sales.csv")

🤖 6 Native Intelligence Agents

1. DataAgent: Universal data loading (CSV, Excel, JSON, Parquet)
2. ProfileAgent: Automatic data profiling and analysis
3. VizAgent: Beautiful HTML dashboards
4. MLAgent: Automatic machine learning
5. DLAgent: Automatic deep learning (PyTorch)
6. InsightAgent: Natural language insights

🚀 What AutoDataMind Does

• Loads Data: CSV, Excel, JSON, Parquet - auto-detected
• Cleans Data: Duplicates, missing values, outliers - automatic
• Analyzes Data: Statistics, correlations, insights - comprehensive
• Visualizes Data: HTML dashboards - professional
• Trains ML Models: Regression, classification - auto-selected
• Trains Deep Models: Neural networks - auto-built
• Generates Reports: Narratives, recommendations - human-readable

📦 Installation

pip install autodatamind

🎓 Quick Start

Analyze Any Dataset

import autodatamind as adm

# Load and analyze - returns complete analysis
analysis = adm.analyze("your_data.csv")

# Access results
print(analysis['overview'])
print(analysis['statistics'])
print(analysis['insights'])

Train ML Model - Zero Code

# Automatic ML training
result = adm.autotrain("sales.csv", target="revenue")

# Get trained model
model = result['model']

# Get metrics
print(result['metrics'])
# {'rmse': 1234.56, 'mae': 987.65, 'r2': 0.89}

# Model saved automatically!

Create Dashboard - One Line

# Generate professional HTML dashboard
adm.dashboard("sales.csv")
# Opens in browser automatically!

Deep Learning - No PyTorch Knowledge

# Automatic deep learning
result = adm.auto_deep(
    "data.csv",
    target="category",
    epochs=50
)

# Get model and metrics
model = result['model']
print(result['metrics'])
# {'accuracy': 0.95}

Get Business Insights

# Generate narrative report
report = adm.generate_insights("sales.csv", target="revenue")

# Report includes:
# - Executive summary
# - Key findings
# - Statistical insights
# - Recommendations
# - Data quality assessment

📊 Complete Example

import autodatamind as adm

# 1. Load data (auto-detected format)
df = adm.read_data("sales.csv")

# 2. Clean data (automatic)
df_clean = adm.autoclean(df)

# 3. Analyze data
analysis = adm.analyze(df_clean)

# 4. Create dashboard
adm.dashboard(df_clean)

# 5. Train ML model
ml_result = adm.autotrain(df_clean, target="revenue")

# 6. Train deep learning model
dl_result = adm.auto_deep(df_clean, target="revenue", epochs=100)

# 7. Generate insights report
report = adm.generate_insights(df_clean, target="revenue")

# Done! 🎉

🎯 Use Cases

Business Analytics

# Analyze sales data
adm.analyze("sales_2024.csv")
adm.dashboard("sales_2024.csv")
adm.generate_insights("sales_2024.csv", target="total_sales")

Predictive Modeling

# Predict customer churn
result = adm.autotrain("customers.csv", target="churn")
print(f"Model accuracy: {result['metrics']['accuracy']:.2%}")

Data Exploration

# Explore new dataset
adm.analyze("new_data.csv")  # Get overview
adm.dashboard("new_data.csv")  # Visual exploration

Report Generation

# Generate executive report
report = adm.generate_insights(
    "quarterly_data.csv",
    target="profit",
    save_report=True
)

🏗️ Architecture

AutoDataMind uses 6 Native Intelligence Agents:

autodatamind/
├── core/               # Core functionality
│   ├── reader.py      # Universal data loader
│   ├── cleaner.py     # Automatic data cleaning
│   ├── utils.py       # Helper functions
│   └── validator.py   # Data validation
├── agents/            # Intelligence agents
│   ├── data_agent.py       # Data handling
│   ├── profile_agent.py    # Analysis & profiling
│   ├── viz_agent.py        # Visualization
│   ├── ml_agent.py         # Machine learning
│   ├── dl_agent.py         # Deep learning
│   └── insight_agent.py    # Narrative generation
└── models/            # ML/DL engines
    ├── auto_ml.py     # AutoML engine
    └── auto_dl.py     # AutoDL engine

💡 Philosophy: Native Intelligence Layer

Traditional Data Science:

# 40 lines of Pandas code
import pandas as pd
df = pd.read_csv("data.csv")
df = df.dropna()
df = df.drop_duplicates()
# ... 35 more lines ...

AutoDataMind:

# 1 line
adm.analyze("data.csv")

No Pandas knowledge required. No ML expertise needed.

🌍 Target Markets

Emerging Economies

• Africa: Kenya, Nigeria, South Africa, Ghana
• Asia: India, Bangladesh, Philippines, Vietnam
• Latin America: Brazil, Mexico, Colombia

User Segments

• Students: Learn data science without complexity
• Small Businesses: No data science budget
• Researchers: Focus on insights, not code
• Analysts: Fast results without programming

🔬 Technical Details

Supported Data Formats

• CSV: Auto-encoding detection (UTF-8, Latin-1, ISO-8859-1)
• Excel: .xlsx, .xls
• JSON: Multiple orientations
• Parquet: High-performance columnar

Auto-Cleaning Features

• Duplicate removal
• Missing value handling (auto/drop/mean/median/mode)
• Type fixing
• Outlier removal
• Data validation

ML Algorithms

• Classification: RandomForest, GradientBoosting, Logistic Regression, KNN, Naive Bayes
• Regression: RandomForest, GradientBoosting, Linear Regression, Ridge, Lasso

DL Architectures

• MLP: Simple multilayer perceptron
• Deep: Deep neural networks (128→64→32)
• Wide: Wide networks (256→128→64)
• Auto-selection based on data size

📈 Performance

Speed

• Small datasets (<10K rows): <1 second
• Medium datasets (10K-100K): <5 seconds
• Large datasets (>100K): <30 seconds

Accuracy

• AutoML: Competitive with manual tuning
• AutoDL: State-of-the-art architectures
• Auto-hyperparameter tuning: GridSearchCV optimization

🤝 Contributing

Contributions welcome! Areas of interest:

• Additional data formats
• More ML algorithms
• Advanced DL architectures
• New visualization types
• Documentation improvements

📄 License

MIT License - see LICENSE file.

👤 Author

Idriss Olivier Bado

• Email: idrissbadoolivier@gmail.com
• GitHub: @idrissbado

🙏 Acknowledgments

Built with:

• pandas: Data manipulation
• scikit-learn: Machine learning
• PyTorch: Deep learning
• matplotlib/seaborn: Visualization

🔗 Links

• PyPI: https://pypi.org/project/autodatamind/
• GitHub: https://github.com/idrissbado/autodatamind
• Documentation: https://github.com/idrissbado/autodatamind#readme
• Issues: https://github.com/idrissbado/autodatamind/issues

---

Made with ❤️ for the global data science community

Democratizing AI, one line of code at a time