Data Analysis and Machine Learning Streamlit Application

This interactive Streamlit application provides a comprehensive platform for data analysis and machine learning. It allows users to load, explore, preprocess, group data, train various machine learning models, and save/load them for future use.

Features

• Data Loading: Easily upload and preview your datasets in Excel (.xlsx) or CSV (.csv) formats.
• Data Exploration: Gain insights into your data with descriptive statistics, data type displays, missing value summaries, correlation matrices, and a variety of interactive visualizations including histograms, bar charts, box plots, scatter plots, and pie charts.
• Data Preprocessing: Prepare your data for modeling with options to:
  • Use the first row as column headers.
  • Handle missing values using methods like dropping rows/columns, mean, median, or most frequent imputation.
  • Encode categorical variables using One-Hot Encoding or Label Encoding.
  • Normalize numerical data using StandardScaler or MinMaxScaler.
• Data Grouping: Aggregate your data by selected columns and apply various aggregation functions (mean, sum, count, min, max).
• Model Training: Train machine learning models with a user-friendly interface:
  • Select target and feature columns.
  • Choose between classification (Logistic Regression, Decision Tree, Random Forest) and regression (Linear Regression, Decision Tree, Random Forest) tasks.
  • Configure the test set size.
  • Optionally perform hyperparameter optimization using GridSearch.
  • Optionally perform cross-validation to assess model robustness.
• Model Evaluation & Visualization: Evaluate trained models with key metrics and visualizations:
  • Display accuracy (for classification) or Mean Squared Error (MSE) and R² score (for regression).
  • Visualize classification results with ROC curves.
  • Visualize regression results with residuals plots.
  • Display feature importance for applicable models.
• Model Saving & Loading: Save your trained models along with their feature and target column information, and load previously saved models for deployment or further analysis.

Technologies Used

• Python
• Streamlit
• Pandas
• NumPy
• Matplotlib
• Seaborn
• Scikit-learn
• Joblib
• Chardet

Installation

To set up and run this project locally, follow these steps:

1. Clone the repository:

   git clone https://github.com/Ysen0603/DataScience_app
   cd DataScience_app

2. Create a virtual environment (recommended):

   python -m venv venv

3. Activate the virtual environment:

   • Windows:

     .\venv\Scripts\activate

   • macOS/Linux:

     source venv/bin/activate

4. Install dependencies:

   pip install -r requirements.txt

Usage

To run the Streamlit application:

1. Activate your virtual environment (if not already active).

2. Navigate to the project root directory (where main.py is located).

3. Run the Streamlit application:

   streamlit run main.py

   This command will open the application in your default web browser.

4. Interact with the application:

   • Use the sidebar navigation to switch between different sections: "Chargement des données" (Data Loading), "Exploration des données" (Data Exploration), "Prétraitement" (Preprocessing), "Groupement" (Grouping), "Entraînement du modèle" (Model Training), and "Sauvegarde du modèle" (Model Saving).
   • Follow the on-screen instructions to upload data, perform analysis, train models, and save/load them.

Project Structure

.
├── main.py                     # Main Streamlit application entry point
├── data/                       # Directory for sample datasets
│   └── Titanic.csv             # Example dataset
├── PG/                         # Contains modules for different stages of the data pipeline
│   ├── __init__.py
│   ├── data_loading.py         # Handles data upload and initial display
│   ├── data_exploration.py     # Provides tools for data visualization and statistics
│   ├── data_preprocessing.py   # Manages data cleaning and transformation
│   ├── data_grouping.py        # Enables data aggregation and grouping
│   ├── model_training.py       # Facilitates machine learning model training
│   └── model_saving.py         # Manages saving and loading of trained models
├── utils/                      # Contains utility functions
│   ├── data_utils.py           # Helper functions for data handling
│   ├── model_utils.py          # Helper functions for model training and evaluation
│   └── preprocessing_utils.py  # Helper functions for data preprocessing
└── requirements.txt            # List of Python dependencies

Contributing

Contributions are welcome! Please feel free to open issues or submit pull requests.

Contact

Developed with ❤️ by Ennaya Yassine yassineennaya@gmail.com