MACHINE LEARNING BASED SOFTWARE BUG PREDICTION SYSTEM

Abstract

In the current world, the usage of mobile applications and suspicious applications has increased rapidly. As the number of applications grows, the risk of privacy breaches and security threats also increases. Therefore, protecting user privacy and identifying suspicious applications has become very important.These defects can reduce software reliability, affect performance, and increase maintenance costs. Therefore, identifying software bugs in the early stages of development is very important to improve software quality and reduce debugging time. Traditional software testing methods mainly depend on manual inspection and rule-based techniques, which are often time-consuming and may not effectively detect hidden defects in large software systems.Machine Learning (ML) techniques provide an efficient solution for predicting software defects. By analyzing historical software data, ML models can learn patterns related to previous bugs and predict whether a software module is likely to contain defects. In this project, several machine learning algorithms such as Support Vector Classifier (SVC), Random Forest (RF), NuSVC, and Multi-Layer Perceptron (MLP) are used to build a software bug prediction model. The system includes different stages such as data collection, preprocessing, feature extraction, model training, and model evaluation. Experimental results show that machine learning models can effectively predict defect-prone modules and help developers focus on critical areas during testing. Among the algorithms used, Random Forest provides better prediction performance compared to other models. The proposed system improves software quality and supports developers in building more reliable and efficient software systems.