Software Bug Prediction

Software quality assurance is a vital aspect of software development, where predicting bugs in code modules before deployment can significantly reduce maintenance costs and improve reliability.This project, Software Bug Prediction, presents a machine learning-based web application built using Flask to predict the likelihood of bugs in software files using static code metrics and project metadata.The backend system integrates two effective classification algorithms-Random Forest Classifier and Logistic Regression-to identify the presence of bugs.It considers features such as module type, programming language, lines of code, cyclomatic complexity, developer experience, code review scores, and test coverage.Categorical features are label encoded, and all features are standardized to enhance model performance.The application trains both models and evaluates them on a hold-out test set, reporting metrics such as accuracy and feature importance.Through a web interface, users can upload or use a dataset, initiate model training, and enter input feature values for prediction.The system outputs the probability of bugs from both models and gives a final result based on an average probability threshold.Additionally, it offers insights like bug rate and feature distribution.This scalable system supports automated bug risk assessment and improves decision-making in software development.