Student feedback plays a crucial role in improving the quality of teaching and learning as well as enhancing training programs. This paper proposes an approach to support the educational sector by automatically analyzing student feedback through their comments. The proposed approach combines the Convolutional Neural Network (CNN) and the Long Short Term Memory (LSTM) in two stages. The first stage for sentiment analysis (e.g., satisfying, unsatisfying) and the second stage for categorizing student feedback into several topics such as their comments about infrastructure, teaching method, learning resources, etc. For experiments, this research uses a data set comprising 5,633 student comments, divided into 64% for training, 16% for validation and 20% for testing. The experimental results show that the proposed method can achieve the F1 scores at 95% for sentiment classification. In topic classification, the F1 scores can get 91.1% for common topics. These results show that the student feedback can be automatically analyzed for improving teaching and learning activities, thus, it helps the university in improving the quality and saving considerable time and effort compared to manual review processes.