Codeforces Rating Predictor

A Codeforces Rating Predictor that estimates changes in user ratings after a contest. The project utilizes machine learning techniques to analyze performance and predict ratings based on contest standings. The prediction model is integrated into a browser extension for seamless interaction with the Codeforces platform.

Features

• Prediction of Rating Changes: Based on user performance and relative standings in a contest.
• Machine Learning Model: Uses Random Forest algorithm optimized with Bayesian optimization.
• Browser Extension: Built using HTML, CSS, and JavaScript, allowing users to easily predict rating changes directly on Codeforces.
• Flask API: A local server processes the data and returns predictions.

Table of Contents

• Project Overview
• Technical Stack
• Data Preprocessing
• Machine Learning Model
• Browser Extension
• Usage
• Setup
• Future Improvements
• License

Project Overview

The aim of this project is to predict the rating changes for Codeforces participants after a contest. By analyzing historical data and the standings from a particular contest, the system estimates the relative performance of each participant and provides an accurate prediction of their new rating.

Technical Stack

• Programming Languages: Python, JavaScript
• Backend: Flask
• Frontend: HTML, CSS, JavaScript
• Machine Learning: Random Forest with Bayesian Optimization
• Data Handling: NumPy, Pandas
• Model Deployment: Flask API on Local Server

Data Preprocessing

The preprocessing step plays a vital role in this project. The following strategies were employed:

1. Surrounding Averages: Compute the average ratings of participants in close ranking positions. This helps determine the relative performance of a participant during a contest.
2. Feature Engineering: Various features were created based on past ratings, ranks, and other relevant metrics.
3. Sorting & Cleaning: Data was sorted based on old ratings, and any irrelevant or missing data was handled accordingly.

Machine Learning Model

Algorithm

• The project uses the Random Forest algorithm for prediction. It is an ensemble learning method that aggregates the predictions from multiple decision trees to improve accuracy and robustness.
• Bayesian Optimization was used to fine-tune the hyperparameters of the model, ensuring that it provides the most accurate predictions possible.

Model Evaluation

• Performance was evaluated using metrics like Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE).
• Cross-validation was conducted to ensure the model generalizes well on unseen data.

Browser Extension

The browser extension was developed using HTML, CSS, and JavaScript. Users can interact with the extension to:

• Select a Codeforces contest.
• Input their handle to get a prediction of their new rating.
• View the results directly within the extension without navigating away from the Codeforces website.

Features of the Extension

• User-Friendly UI: Simple and intuitive design.
• Real-Time Predictions: Fetches contest data and sends it to the local server for processing.
• Integration with Codeforces: Allows users to see predictions based on live data from the platform.

Usage

1. Start the Flask Server:

   python app.py

2. Use the Browser Extension:

   • Install the extension by loading the manifest.json file in your browser.
   • Navigate to a Codeforces contest page.
   • Open the extension and input your handle.
   • Click on "Predict" to see the rating change estimate.

Setup

Prerequisites

• Python 3.x
• Flask
• scikit-learn
• Pandas
• NumPy
• Bayesian Optimization Libraries: e.g., scikit-optimize
• Browser with Extension Support: Chrome, Firefox, etc.

Step-by-Step Setup

1. Clone the Repository

   git clone https://github.com/yourusername/CF-Rating-Predictor.git
   cd CF-Rating-Predictor

2. Install Python Dependencies

   pip install -r requirements.txt

3. Run the Flask API

   python app.py

4. Install the Browser Extension

   • Go to chrome://extensions/ (or the equivalent for your browser).
   • Enable "Developer mode".
   • Click on "Load unpacked" and select the extension/ directory.

5. Make Predictions: Open the extension, select a contest, input your handle, and get predictions!

Future Improvements

• Model Enhancements: Integrate more sophisticated algorithms like Gradient Boosting Machines (GBM) or neural networks for better accuracy.
• Data Source Automation: Automate the fetching of historical data to update the model regularly.
• Deployment: Deploy the Flask API on a cloud server to make it accessible anywhere.
• Cross-Browser Compatibility: Extend the extension support to other popular browsers.

License

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

---