Student Behavior Detection on Campus using Ghost Crayfish Optimization Algorithm and Bi Directional Long Short-Term Memory

Student engagement is a supple and dynamic concept encompassing behavioural and cognitive involvement. To help instructors better understand how student engage with several classroom activities, it is crucial to predict their level of participation. However, were difficult to interpret, making it challenging for educators or administrators to inefficiently learn pattern. This paper, proposed Ghost Crayfish Optimization Algorithm (GCOA) for feature selection and Bi Directional Long Short-Term Memory (Bi-LSTM) for student behavior classification ensures better accuracy. The GCOA effectively reduce the dimensionality of data, selecting only most relevant feature while Bi-LSTM technique processing sequential data, handle long-term dependencies and gradient vanishing issue. Initially, data is obtained from Open University Learning Analytics Dataset (OULAD) dataset and pre-processing stage involves missing values, normalization which is efficiently handle missing values and scale feature. The proposed GCOA and Bi-LSTM technique achieves better accuracy 94.26%, precision of 94.75% and recall of 94.12% on OULAD dataset when compared to existing techniques Artificial Neural Network (ANN), Fully Connected Network (FCN) and Long Short-Term Memory (LSTM) approach.