ML Lab

A comprehensive desktop application for exploring datasets, training machine learning models, and evaluating their performance. Built with Python and Flet for a beautiful, cross-platform UI.

Features

• Dataset Management

  • Load and explore CSV files with intuitive data browsing
  • View dataset statistics and summary information
  • Edit individual cell values directly in the table
  • Rename and drop columns
  • Handle missing values (detect NaN rows, drop duplicates)
  • Export modified datasets with timestamps
  • Filter and search by row index or range
  • Replace values in selected columns
  • Change data types of columns

• Data Visualization

  • Interactive charts: scatter plots, histograms, bar charts, box plots, pie charts, heatmaps
  • Customizable visualizations based on selected columns
  • Real-time chart generation

• Machine Learning Models

  • Classification Models: Logistic Regression, Decision Tree, Random Forest, SVM, KNN, Gradient Boosting, Naive Bayes models, Extra Tress, Ada Boost XGBoost, LightGBM, CatBoost, SGD, MLP (Neural Network)
  • Regression Models: Linear Regression, Decision Tree Regressor, Random Forest, Ridge Regression, Lasso Regression, ElasticNet, KNN, Gradient Boosting, Extra Trees, AdaBoost, XGBoost, LigtGBM, CatBoost, SVM, SGD, MLP (Neural Network)
  • Clustering Models: K-Means, DBSCAN, Hierarchical Clustering, Gaussian Mixture Models, Mean Shift, HDBSCAN, Affinity Propagation, MiniBatch K-Means, OPTICS, Spectral Clustering, Birch
  • Helper Tools: Elbow locator for optimal cluster selection

• Model Configuration

  • Flexible hyperparameter tuning with validation
  • Multiple scaling options: Standard Scaler, MinMax Scaler, None
  • Features encoder options: OrdinalEncoder, TargetEncoder, OneHotEncoder, LabelEncoder
  • Target encoder options: None, LabelEncoder
  • Configurable train/test split for supervised learning
  • Automatic feature selection
  • Test new data on trained model

• Model Evaluation

  • Classification metrics: Accuracy, Precision, Recall, F1-Score, Confusion Matrix
  • Regression metrics: R² Score, Adjusted R² Score, Mean Squared Error, Mean Absolute Error, Mean Absolute Percentage Error
  • Clustering metrics: Silhouette Score, Davies-Bouldin Index, Calinski-Harabasz Index
  • Detailed results with copy functionality
  • Provide a code block to recreate model

• Data Profiling

  • YData Profiling integration for comprehensive dataset reports
  • Export HTML reports for detailed analysis

Installation

Using uv (Recommended)

First, ensure you have uv installed.

# Clone the repository
git clone https://github.com/farshadz1997/ML-Lab.git
cd "ML-Lab"

# Create virtual environment & Install dependencies
uv sync

# Activate virtual environment
# On Windows:
.venv\Scripts\activate
# On macOS/Linux:
source .venv/bin/activate

cd src

# Run the app as a desktop application
uv run main.py

Using pip

# Clone the repository
git clone https://github.com/farshadz1997/ML-Lab.git
cd "ML-Lab"

# Create a virtual environment (recommended)
python -m venv venv

# Activate virtual environment
# On Windows:
venv\Scripts\activate
# On macOS/Linux:
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

cd src

# Run the app as a desktop application
python main.py

Quick Start

1. Load a Dataset: Click "Open Dataset" and select a CSV file
2. Explore Data: Browse the dataset, view statistics, check for missing values
3. Select Model: Navigate to the Model Training tab
4. Configure Hyperparameters: Adjust model settings as needed
5. Train & Evaluate: Click "Train and evaluate model" to see results
6. Visualize: Use the Data Visualization tab to create charts

Requirements

• Python 3.11+
• See requirements.txt for all dependencies

Building for Distribution

Windows

flet build windows -v

For more details on building Windows package, refer to the Windows Packaging Guide.

Project Structure

src/
├── main.py                 # Application entry point
├── core/
│   └── dataset.py         # Dataset handling and operations
│   └── data_preparation.py          # Main encoding orchestration
│       ├── prepare_data_for_training()       - Complete data prep pipeline
│       └── prepare_data_for_training_no_split() - Clustering variant
├── ui/
│   ├── layout.py          # Main UI layout
│   ├── data_visualization.py
│   ├── dataset_explorer.py # Dataset exploration interface
│   ├── model_factory.py    # Model training interface
│   ├── charts/            # Chart implementations
│   └── models/            # ML model wrappers
├── utils/
│    └── model_utils.py     # Model utilities and metrics
│       ├── detect_categorical_columns()
│       ├── validate_cardinality()
│       ├── create_categorical_encoders()
│       ├── apply_encoders()
│       ├── get_encoding_mappings()
│       ├── build_preprocessing_pipeline()
│       ├── compose_full_model_pipeline()
│       ├── get_categorical_encoding_info()
│       └── Data classes: EncodingError, CardinalityWarning, CategoricalEncodingInfo

License

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

Contributing

Contributions are welcome! Feel free to submit issues and pull requests.