House Price Prediction Model

This project is a comprehensive machine learning solution that predicts house prices based on various features. The project follows MLOps principles, including CI/CD pipelines, experiment tracking, and orchestration, to ensure efficient and reproducible workflows. It also implements software design patterns to make the codebase scalable and maintainable.

Features

• Machine Learning: Implements various ML algorithms for house price prediction.
• MLOps Integration: Includes CI/CD pipelines for continuous model development and deployment.
• Experiment Tracking: Utilizes MLflow to track experiments, models, and hyperparameters.
• Pipeline Orchestration: ZenML is used to orchestrate and deploy ML pipelines.
• Design Patterns: The code follows best practices and design patterns to ensure modularity and scalability.

Tech Stack

• Python 3.8+
• MLflow: For experiment tracking.
• ZenML: For pipeline orchestration and model deployment.
• Pandas, Scikit-learn: For data preprocessing and model development.
• TensorFlow: For deep learning models.
• Flask: (Optional) For serving the model as a web service.
• Docker & Kubernetes: (Optional) For containerization and deployment at scale.

Installation and Setup

Prerequisites

• Python 3.8+
• Git
• Virtual environment tools (venv or virtualenv)
• ZenML and MLflow integrations

Step-by-Step Setup

1. Clone the Repository

   First, clone the project repository from GitHub:

   git clone https://github.com/Madhavyamjala/house-price-prediction.git
   cd house-price-prediction

2. Create a Virtual Environment

   Follow this guide to create a virtual environment: Create Virtual Environment Guide

   Once the virtual environment is activated:

   source venv/bin/activate  # For Linux/macOS
   venv\Scripts\activate     # For Windows

3. Install Required Dependencies

   Install all necessary Python libraries and dependencies:

   pip install -r requirements.txt

4. Install ZenML and MLflow Integrations

   The project uses ZenML for pipeline orchestration and MLflow for tracking experiments. Install the necessary integrations:

   zenml integration install mlflow -y

5. Configure the ZenML Stack

   You need to configure the ZenML stack with the experiment tracker and model deployer:

   zenml experiment-tracker register mlflow_tracker --flavor=mlflow
   zenml model-deployer register mlflow --flavor=mlflow
   zenml stack register local-mlflow-stack -a default -o default -d mlflow -e mlflow_tracker --set

6. Run the ZenML Pipeline

   To execute the ML pipeline, use the run_pipeline.py script:

   python run_pipeline.py

7. Deploy the Model

   If you'd like to deploy the model, use the run_deployment.py script:

   python run_deployment.py

Project Structure

├── analysis            # Scripts and notebooks for data analysis
├── data                # Raw and processed dataset files
├── explanations        # Explanatory documentation or data insights
├── extracted_data      # Extracted data from preprocessing steps
├── mlruns              # MLflow run tracking artifacts
├── pipelines           # ZenML pipelines for the project
├── src                 # Source code for the model and utilities
├── steps               # Individual steps in the ZenML pipeline
├── tests               # Unit and integration tests for the codebase
├── config.yaml         # Configuration file for the project
├── requirements.txt    # Python dependencies
├── run_deployment.py   # Script for deploying the model
├── run_pipeline.py     # Script for running the ZenML pipeline
├── sample_predict.py   # Script for running a sample prediction

Experiment Tracking with MLflow

All experiments are tracked using MLflow, which logs:

• Model artifacts
• Hyperparameters
• Performance metrics (e.g., accuracy, loss)

You can visualize the experiment results using the MLflow UI:

mlflow ui

Orchestrating Pipelines with ZenML

ZenML orchestrates the ML pipelines and handles tasks such as data ingestion, training, evaluation, and deployment. You can modify the steps of the pipeline inside the pipelines and steps directories.

Testing

Unit tests and integration tests are available in the tests folder. To run the tests, use the following command:

pytest tests/

Contributing

Feel free to fork this repository and submit a pull request if you'd like to contribute. All contributions are welcome.

License

This project is licensed under the MIT License. See the LICENSE file for more details.