Software Fault Prediction using Advanced Deep Learning

This project uses advanced deep learning techniques to predict software faults. Dataset: The dataset used for this project contains information about software faults. Principal Component Analysis (PCA) is used to reduce the dimensionality of the dataset. The Adasyn method is then applied to handle the class imbalance problem. Model: The model used for this project is a combination of Gated Recurrent Unit (GRU) with Long Short-Term Memory (LSTM) architecture. The first layer is Bi-LSTM, the second layer is Bi-GRU, and the third layer is LSTM, with each node having 150 nodes. The attention mechanism is used to generate important features and reduce complexity to improve accuracy. Results: The results of this project show that the advanced deep learning model is effective at predicting software faults. The attention mechanism was able to identify important features and reduce complexity, resulting in improved accuracy