Automated Code Efficiency Evaluation and Ranking System using Machine Learning

The importance of performance and correctness evaluations in automated assessment systems and competitive programming now holds an important place. In fact, the traditional evaluation systems have evaluated correctness without taking into account speed, memory usage, and code complexity. This paper presents an AI-based model that applies machine learning techniques to predict and rank C++ code submissions.Features such as LOC, Cyclomatic Complexity, Execution Time, and Memory Usage are extracted from the Project Codenet dataset. An efficiency score is calculated on the basis of a weighted formula combining structural complexity and runtime resource consumption. A Random Forest Regressor is trained on these scores to predict their values for high accuracy in submission ranking.To make it more user-friendly, the system has an interactive web-based UI which visualizes submission ranking, efficiency scores, and performance metrics. It allows filtering by problem ID, showcasing graphical trends, and comparing code efficiency to make it easier to find better quality solutions. This system allows for real-time assessment with minimal manual effort for scaling to large-scale coding environments.The framework can not only automate the assessment of the code but also act as a teacher for programmers and teachers encouraging them to code effectively. Future work will include using deep learning models for prediction accuracy, alongside containerized environments (like Docker) to give more precise benchmark results. The proposed system will have wide-ranging applications in programming training, coding competitions, and platform evaluation strategies.