TY  - JOUR
AU  - Wu, Meiting
PY  - 2026
DA  - 2026/03/07
TI  - K-Means Clustering-Based Feature Generation for Student Performance Prediction
JO  - ICCK Transactions on Educational Data Mining
T2  - ICCK Transactions on Educational Data Mining
JF  - ICCK Transactions on Educational Data Mining
VL  - 2
IS  - 1
SP  - 14
EP  - 28
DO  - 10.62762/TEDM.2026.716076
UR  - https://www.icck.org/article/abs/TEDM.2026.716076
KW  - clustering feature generation
KW  - academic performance prediction
KW  - educational data mining
KW  - machine learning
KW  - clustering algorithm
AB  - With the development of educational technology and the accumulation of big data, student performance prediction has become a hot topic in the field of education. However, traditional manual statistical methods have limitations in dealing with complex data and are difficult to achieve high-precision prediction. To address this gap, this study proposes a clustering-based feature generation framework to enhance prediction performance. Firstly, the multilayer perceptron (MLP) model is employed to evaluate the effectiveness of the clustering algorithms (K-Means, DBSCAN, and hierarchical clustering) for feature generation. Then, the best clustering algorithm (K-Means) is applied to generate features that are subsequently integrated with original features to construct augmented datasets. Subsequently, grid search is adopted to optimize model hyperparameters, and six machine learning models, including MLP, support vector machine, random forest, Bagging, XGBoost, and CatBoost, are trained and evaluated on datasets with and without clustering-generated features. Experimental results demonstrate that K-Means-based feature generation can effectively improve prediction performance under certain conditions. However, the performance gains are influenced by data characteristics, feature distributions, and model structures. The findings also reveal that clustering-derived features do not universally enhance all machine learning algorithms, highlighting the necessity of selecting appropriate model–feature integration strategies in practical applications.
SN  - 3070-5843
PB  - Institute of Central Computation and Knowledge
LA  - English
ER  - 

@article{Wu2026KMeans,
  author = {Meiting Wu},
  title = {K-Means Clustering-Based Feature Generation for Student Performance Prediction},
  journal = {ICCK Transactions on Educational Data Mining},
  year = {2026},
  volume = {2},
  number = {1},
  pages = {14-28},
  doi = {10.62762/TEDM.2026.716076},
  url = {https://www.icck.org/article/abs/TEDM.2026.716076},
  abstract = {With the development of educational technology and the accumulation of big data, student performance prediction has become a hot topic in the field of education. However, traditional manual statistical methods have limitations in dealing with complex data and are difficult to achieve high-precision prediction. To address this gap, this study proposes a clustering-based feature generation framework to enhance prediction performance. Firstly, the multilayer perceptron (MLP) model is employed to evaluate the effectiveness of the clustering algorithms (K-Means, DBSCAN, and hierarchical clustering) for feature generation. Then, the best clustering algorithm (K-Means) is applied to generate features that are subsequently integrated with original features to construct augmented datasets. Subsequently, grid search is adopted to optimize model hyperparameters, and six machine learning models, including MLP, support vector machine, random forest, Bagging, XGBoost, and CatBoost, are trained and evaluated on datasets with and without clustering-generated features. Experimental results demonstrate that K-Means-based feature generation can effectively improve prediction performance under certain conditions. However, the performance gains are influenced by data characteristics, feature distributions, and model structures. The findings also reveal that clustering-derived features do not universally enhance all machine learning algorithms, highlighting the necessity of selecting appropriate model–feature integration strategies in practical applications.},
  keywords = {clustering feature generation, academic performance prediction, educational data mining, machine learning, clustering algorithm},
  issn = {3070-5843},
  publisher = {Institute of Central Computation and Knowledge}
}