🤖 Advanced AutoML CSV Evaluator

An intelligent, end-to-end machine learning automation tool that handles any kind of CSV data with automatic cleaning, preprocessing, model selection, and evaluation.

✨ Key Features

🧹 Automatic Data Cleaning

• Duplicate removal - Identifies and removes duplicate rows
• Missing value handling - Fills categorical with mode, numerical with median
• Outlier detection - Uses IQR method to clip outliers in numerical columns
• Infinite value handling - Replaces inf/-inf with median values
• Data type detection - Automatically identifies categorical vs numerical columns

🤖 Smart Problem Detection

• Automatic Classification Detection - Detects categorical targets & discrete values (< 20 unique)
• Automatic Regression Detection - Identifies continuous numerical targets
• Adaptive Model Selection - Chooses appropriate models based on problem type

📊 Comprehensive Model Library

Classification Models:

• Logistic Regression
• Random Forest Classifier
• Gradient Boosting Classifier
• Support Vector Machines (SVM)
• K-Nearest Neighbors
• Decision Tree Classifier
• Naive Bayes

Regression Models:

• Linear Regression
• Ridge Regression
• Lasso Regression
• Random Forest Regressor
• Gradient Boosting Regressor
• Support Vector Regression (SVR)
• K-Nearest Neighbors Regressor
• Decision Tree Regressor

📈 Advanced Evaluation

• Cross-Validation (5-fold) - Robust model evaluation with CV scores
• Multiple Metrics
  • Classification: Accuracy, Precision, Recall, F1-Score, CV metrics
  • Regression: R² Score, MAE, MSE, RMSE, CV metrics
• Visual Comparisons - Interactive Plotly charts comparing model performance
• Best Model Selection - Automatically identifies and highlights the best performing model

💾 Export & Reporting

• Download results as CSV
• Detailed summary reports with dataset information
• Feature and target variable analysis
• Actionable recommendations

🌐 Live Demo

Try it now! The app is deployed and ready to use:

No installation required — just upload your CSV and start analyzing! The live demo includes:

• ✅ Full functionality (data upload, cleaning, model training)
• ✅ Light & Dark mode support
• ✅ Sample datasets (iris.csv, air.csv) included
• ✅ Instant results and visualizations

🚀 Quick Start

Prerequisites

• Python 3.8 or higher
• pip or conda package manager

Installation (Local)

Option 1: Using pip

# Clone the repository
git clone https://github.com/Pujan-Dev/AutoML.git
cd AutoML

# Install dependencies
pip install -r requirements.txt

Option 2: Using conda

conda create -n automl python=3.9
conda activate automl
pip install -r requirements.txt

Running the App

streamlit run main.py

The app will automatically open at http://localhost:8501 in your browser.

Running with Docker

# Build the Docker image
docker build -t automl:latest .

# Run the container
docker run -p 8501:8501 automl:latest

Then visit http://localhost:8501 in your browser.

📋 How to Use

Step-by-Step Guide

1. Upload CSV

   • Use the sidebar file uploader to select your CSV file
   • Supports any tabular CSV format

2. Preview Data

   • View dataset information (rows, columns, missing values, data types)
   • Expand the "Dataset Preview" section to see sample rows

3. Configure Data Options

   • For large datasets (>1000 rows), choose to sample or select columns
   • Remove unnecessary columns from analysis

4. Select Target Column

   • Choose the column you want to predict
   • The app automatically detects classification vs regression

5. Run AutoML

   • Click the "🚀 Run AutoML" button to start training
   • Watch the progress as models are trained sequentially

6. Review Results

   • See model comparison table with all metrics
   • View performance visualization chart
   • Identify the best model (highlighted with 🏆)

7. Export Results

   • Download detailed results as CSV
   • View comprehensive summary report with recommendations

� Screenshots

Main Interface

Clean and intuitive interface for uploading and configuring your data

📊 Sample Datasets

sample datasets are :

• iris.csv - Classification (predicting flower species)
• air.csv - Regression (predicting air quality metrics)

🔧 What Happens Under the Hood

Upload CSV
    ↓
🧹 Auto Clean Data (duplicates, missing values, outliers)
    ↓
🔍 Detect Problem Type (Classification vs Regression)
    ↓
⚙️ Build Preprocessing Pipeline
    • Impute numerical features (median)
    • Impute categorical features (mode)
    • One-hot encode categorical variables
    • Scale numerical features
    ↓
🤖 Train Multiple Models (7-8 models depending on task type)
    • 5-fold Cross-Validation for each model
    • Full training set fitting
    ↓
📊 Evaluate on Test Set
    • Calculate metrics (Accuracy/Precision/Recall/F1 for classification)
    • Calculate metrics (R²/MAE/MSE/RMSE for regression)
    ↓
🏆 Select Best Model & Display Results
    • Model comparison table
    • Performance visualization
    • Detailed report generation
    ↓
💾 Export Results (CSV download available)

📝 Example Workflow

Classification Example (Iris Dataset):

• Upload: iris.csv
• Target: species
• Auto-detected: Classification
• Models trained: 7
• Best model: Random Forest (98.3% accuracy)
• Metrics: Accuracy, Precision, Recall, F1-Score

Regression Example (Air Quality Dataset):

• Upload: air.csv
• Target: AQI_value
• Auto-detected: Regression
• Models trained: 8
• Best model: Gradient Boosting (R² = 0.92)
• Metrics: R², MAE, MSE, RMSE

🛡️ Error Handling

The app gracefully handles:

• Missing values in any column
• Mixed data types (strings, numbers, booleans)
• Datasets with too few or too many samples
• Categorical variables with high cardinality
• Models that fail to train (skips with warning)
• Infinite and NaN values

📦 Dependencies

All dependencies are listed in requirements.txt:

streamlit>=1.28.0
pandas>=1.5.0
numpy>=1.23.0
scikit-learn>=1.3.0
plotly>=5.14.0

For development, install with:

pip install -r requirements.txt

Version Compatibility

• Python: 3.8, 3.9, 3.10, 3.11
• Streamlit: 1.28.0+
• scikit-learn: 1.3.0+
• pandas: 1.5.0+
• numpy: 1.23.0+
• plotly: 5.14.0+

🎯 Use Cases

• Quick model prototyping - Test multiple algorithms rapidly
• Data exploration - Understand which models work best for your data
• Baseline establishment - Get baseline results before fine-tuning
• Non-technical users - No ML expertise needed
• Competition prep - Quick EDA and model benchmarking
• Production POC - Validate model viability quickly

🔮 Advanced Features

• Automatic problem type detection
• Cross-validation for robust evaluation
• Missing data handling (statistical imputation)
• Categorical encoding (one-hot encoding)
• Feature scaling (StandardScaler)
• Outlier detection and handling
• Parallel model training (n_jobs=-1)
• Interactive visualizations

📄 Output Format

The CSV results file contains:

• Model name
• All performance metrics
• Cross-validation mean and std
• Easy comparison across models

Example:

Model,Accuracy,Precision,Recall,F1 Score,CV Mean,CV Std
Logistic Regression,0.9667,0.9667,0.9667,0.9667,0.9667,0.0211
Random Forest,0.9833,0.9833,0.9833,0.9833,0.9833,0.0178
...

🎨 UI Features

• Light & Dark Mode - Toggle between light and dark themes in the sidebar under "Appearance"
• Responsive Design - Works seamlessly on desktop, tablet, and mobile browsers
• Interactive Charts - Hover over visualizations for detailed metrics
• Real-time Updates - Live progress indicators during model training
• Exportable Results - Download analysis results in CSV format

⚙️ Configuration

Theme Selection

In the sidebar under "Appearance", you can toggle between:

• Light Mode - Clean, bright interface for daytime use
• Dark Mode - Easy on the eyes for extended sessions

⚠️ Limitations

• Currently optimized for tabular CSV data
• Time series and sequential data need preprocessing
• Image and text data not supported (use specialized models)
• Very large datasets (>100k rows) may be slow
• Categorical columns with >1000 unique values may cause memory issues

🚀 Future Enhancements

• Hyperparameter tuning with Bayesian optimization
• Feature importance analysis
• SHAP value explanations
• Time series specialized models
• Ensemble model creation
• Model persistence and loading
• Prediction on new data
• Automated feature engineering
• Class imbalance handling
• GPU support for large datasets

📞 Support & Contact

For issues or questions, please open an issue in the repository.

---

🔗 Links

• GitHub: github.com/Pujan-Dev
• Portfolio: neupanepujan.com.np

---

Made with ❤️ for making AutoML accessible to everyone!