Predict productivity loss by analyzing social media usage patterns! This project explores how social media usage impacts productivity and provides a predictive model to assess productivity loss. It combines data exploration, preprocessing, model training, and an interactive web app interface for a smooth user experience.

🔗 Live Demo: Productivity Loss Predictor
🔗 GitHub Repo: Productivity Loss Predictor
📊 Data Source: Time Wasted on Social Media - Kaggle

• Predict Productivity Loss: Model the impact of social media use on productivity using machine learning.
• Extensive Data Processing: Clean, preprocess, and transform data for accurate prediction.
• Multiple Algorithms Tested: Tested both regression and classification algorithms to find the best model for predicting productivity loss.
• Voting Ensemble for Accuracy: Top 3 models are combined using a hard voting ensemble to maximize prediction accuracy.
• Streamlit Web Application: Provides an intuitive UI for end-users to make predictions and visualize results.
• Dockerized Deployment: Easily deployable Docker setup for streamlined application hosting.

1. Requirements Installation: Install necessary packages from dev.requirements.txt and prod.requirements.txt.
2. Data Exploration: Analyze dataset features and distributions.
3. Data Cleaning: Handle missing data, remove duplicates, and filter outliers.
4. Data Preprocessing: Normalize numerical data, encode categorical data, and perform feature scaling.
5. Model Preparation: Train and evaluate multiple models to identify the top performers.
6. Model Selection and Pipeline Creation: Assemble the top models into a voting classifier.
7. Pipeline Testing: Ensure accuracy and reliability of the entire model pipeline.
8. Model Exporting: Save the final model and unique values using pickle for easy deployment.

To create a robust prediction model, several machine learning algorithms were tested, both in regression and classification modes. The final model is a voting classifier, which leverages the strengths of the top 3 models identified during testing.

Regression Algorithms:

• Linear Regression
• Lasso CV (L1 regularization to reduce overfitting)
• Ridge CV (L2 regularization)
• Support Vector Machine Regression

Classification Algorithms:

• Decision Tree Classifier
• Random Forest Classifier
• Support Vector Machine Classifier
• K-Neighbors Classifier
• Gradient Boosting Classifier

Based on accuracy metrics (especially the r2 score), the top 3 models chosen for the ensemble were:

• Decision Tree Classifier
• Gradient Boosting Classifier
• Random Forest Classifier

These models were combined using a hard voting classifier, enabling the final model to leverage their combined strengths.

data/
  ├── time-wasted-on-social-media.csv       # Dataset used for training

output/
  ├── model_pipeline.pkl                    # Exported model pipeline including preprocessor and model
  ├── model_unique_values.pkl               # Saved unique values for consistent data processing

webapp/
  ├── __init__.py
  ├── app.py                                # Streamlit app for UI and user interaction

.dockerignore
.gitignore
Dockerfile
docker-compose.yml
requirements/
  ├── dev.requirements.txt                  # Development dependencies
requirements.txt                            # Optimized production dependencies

training.ipynb                              # Jupyter Notebook with training pipeline and model exploration

• Dockerfile: Configured for streamlined deployment of the Streamlit app with all dependencies.
• Exposes Port: 8501 (Access the app on localhost:8501)

• Python 3.8+
• Docker
• Streamlit

1. Clone the repository:

   git clone https://github.com/ketanip/productivity-loss-predictor.git
   cd productivity-loss-predictor

2. Install dependencies: For development:

   pip install -r requirements/dev.requirements.txt

   For production:

   pip install -r requirements/prod.requirements.txt

3. Run the app locally:

   streamlit run webapp/app.py

4. Docker Deployment: Build and run the Docker container:

   docker build -t productivity-loss-predictor .
   docker run -p 8501:8501 productivity-loss-predictor

docker compose up --build

Contributions are welcome! To contribute:

1. Fork the repository.
2. Create a feature branch (git checkout -b feature-branch).
3. Commit your changes (git commit -m 'Add new feature').
4. Push to the branch (git push origin feature-branch).
5. Open a pull request.

This project is licensed under the MIT License.

For questions or issues, please open an issue in the repository or contact the project maintainer through GitHub.