Research on Anomaly Detection Algorithm for Course Assessment Results Based on Data Mining

This paper proposes an anomaly detection method for course assessment results combined with the variational autoencoder (VAE) algorithm. This method identifies potential abnormal data points by analyzing the distribution characteristics of students' scores in course assessments, so as to timely detect abnormal situations of student grade fluctuations, such as cheating and grading errors. First, this paper preprocesses the data, including removing missing values and standardizing. Then, the variational autoencoder is used for feature extraction. The VAE model uses the encoder to compress high dimensional data to low dimension latent space, and then reconstructs the raw data by decoding. The paper can detect the data points which deviate significantly from the model. The experimental results show that the proposed algorithm can effectively identify abnormal data, and has higher accuracy and lower false positive rate than traditional methods. The variational autoencoder algorithm has good stability and accuracy when processing large-scale data. The evaluation indicators such as precision, recall rate and F1 value in the experiment all show the superiority of this method in anomaly detection, especially when dealing with large fluctuations in student grades, it shows a strong ability to identify anomalies.