Quality Evaluation of Talent Cultivation in E-Commerce Major using Deep Neural Networks

In recent years, the rapid advancement of digital technologies, cloud computing, and mobile devices has transformed E-commerce into integral component of global economies. However, the existing Feedforward Neural Network (FNN) model with K-Mode’s clustering algorithm faced challenges on capturing non-linear relationships between academic and student behavioral information, which influences the quality evaluation. Initially, both academic and behavioral data are obtained from Student Performance & Behavior (SPB) dataset. In addition, teaching factors such as teaching factors are collected from Teaching Assistant Evaluation (TAE) dataset. Then, min-max normalization is utilized for numerical data processing and One-Hot Encoding (OHE) technique is used for categorical data which making it more effective for classification. Then, the Deep Neural Network (DNN) model is employed with fully connected layers to classify talent quality into high, moderate, and low categories, which making the model effective for evaluation by capturing non-linear relationships. After that, predicted outputs are optimized through cross-entropy loss function to evaluate quality of talent cultivation in E-commerce major. The experimental results show that, proposed DNN model have excellent performance when compared with existing FNN with K-Modes’s clustering in term of accuracy (82.54 %), and Mean Squared Error (MSE) (49.57%) respectively.