Resampling Technique for Imbalanced Class Handling on Educational Dataset

Educational data mining is an emerging field in mining. The need for accurate identifying student accomplishment on a course or maybe upcoming can help the institution to build technology-aided education better. becoming more important be studied because of its potential produce knowledge base model even teacher lecturer. Like another classification task, educational has common and frequently discovered problem. problem that occurred specifically tasks generally imbalanced class An condition where distribution each not same proportion. In this research, it found severely multiclass dataset consists than two labels. According stated beforehand, paper will focus handling with several methods both such as Linear Regression, Random Forest Stacking SMOTE, ADASYN, SMOTE-ENN resampling algorithm. are being evaluated using 10-fold cross-validation 80-20 splitting ratio. result shows best performance coming from ADASYN resampled ratio 0.97 F1 score. study also technique improves performance. Even though no-resampling produced decent too, caused by things general pattern already been good start. Thus, there no real drawbacks if original processed.