Skip to content

Latest commit

 

History

History
28 lines (21 loc) · 3.06 KB

README.md

File metadata and controls

28 lines (21 loc) · 3.06 KB
Raw

Introduction

The Titanic StreamLit Website is an interactive web platform showcasing machine learning models developed for the Kaggle Titanic dataset. The website features a homepage and dedicated pages for Neural Network, Random Forest, and Gradient Boosted Trees models. This project serves as a testament to the deployment of machine learning models to the cloud, leveraging Streamlit for seamless integration.

Website Structure

  1. Home Page: The homepage (index.html) introduces the project, providing a glimpse into the machine learning models used and the data preprocessing techniques. Links to GitHub repositories and other model pages are also available.

  2. Model Pages:

    • Neural Network (Neural_Network.py): This page, structured with HTML (Neural_Network_1.html and Neural_Network_2.html), presents the neural network model. Users can input custom values like age and gender to receive survival predictions.
    • Random Forest (Random_Forest.py): Similar in structure to the Neural Network page, this page (Random_Forest_1.html and Random_Forest_2.html) focuses on the Random Forest model, allowing users to interact and receive predictions.
    • Gradient Boosted Trees (Gradient_Boosted.py): Showcasing the Gradient Boosted Trees model, this page (Gradient_Boosted_1.html and Gradient_Boosted_2.html) also allows for interactive user input and predictions.

Technical Details

  • Data Preprocessing: Separate preprocessing scripts for each model (data_preprocessing_neural_network.py, data_preprocessing_Random_Forest.py, data_preprocessing_Gradient_Boosted.py) ensure data is formatted correctly for model input.
  • Model Deployment: Models are saved and deployed using Streamlit. For instance, model_12_saved.h5 represents a saved neural network model, and model3_GB/saved_model.pb for the Gradient Boosted Trees.
  • Requirements: The requirements.txt file lists all the necessary libraries and dependencies required to run the models and the website.

Machine Learning Pipeline

  • Model Training: Detailed training and fine-tuning of models are evident in the code and HTML descriptions, ensuring optimal performance.
  • Evaluation and Scoring: The project boasts a high score of 80.622 in the Kaggle competition, indicating the efficacy of the models and the engineering pipeline.

Learnings and Achievements

  • This project illustrates the integration of HTML with Streamlit, displaying the capability to create interactive web applications for machine learning models.
  • It highlights the deployment of models to the cloud, allowing for real-time interaction and prediction based on user input.
  • The repository serves as an educational tool for understanding model deployment, cloud integration, and the use of Streamlit in machine learning projects.

Links and References

For an in-depth view and understanding of this project, visit the GitHub Repository and explore the various HTML and Python files that constitute the website's structure and functionality.