Classification via Clustering for Subject-based Scientific Fields in Kindergarten Students

The concern of teachers and parents to maximize the potential of their children’s golden years is a common phenomenon. Exploring this potential is usually based on a prominent child’s scientific field. This is certainly not easy to do because of the many subjects taken when children attend early childhood education. This problem is crucial and needs to be solved. However, research related to this has not been done much. Therefore, this study focuses on classification via clustering the scientific fields of kindergarten children based on the subjects. The clustering task is done using the K-means method. Then, the classification task is applied by using 2 methods, namely: Random Forest and Support Vector Classifier (SVC). The implementation of these methods to obtain optimal model performance. The experimental results show that the highest performance model can be generated by the proposed method. This is indicated by the optimal model, in terms of the accuracy level, which achieved about 100%. This best performance of the model is built by the K-means-SVC method with a training data size of 80% and 85%.