Anomaly detection to evaluate in-class learning process using distance and density approach of machine learning

This study explores k-Nearest Neighbor (k-NN) on distance approach and local outlier factor (LOF) on density approach for anomaly detection in an in-class evaluation scores dataset. The dataset used for this study is class evaluation scores dataset from Character Building courses performed by a Junior High School in Salatiga, Central Java. The experimentation results show LOF based with density approach outperforms accuracy of k-NN with distance approach in detecting anomalous data from the input dataset.