EVALUATION OF THE K-NEAREST NEIGHBOR MODEL WITH K-FOLD CROSS VALIDATION ON IMAGE CLASSIFICATION

M. Rhifky Wayahdi (Universitas Battuta)
Dinur Syahputra (Universitas Battuta)
Subhan Hafiz Nanda Ginting (Universitas Battuta)

INFOKUM
Vol. 9 No. 1,Desember (2020): Data Mining, Image Processing,artificial intelligence, networking

In this paper, the data used is the banana image which is extracted into the dataset into 4 attributes, namely red, green, blue, and the mean for the classification process. Image data is classified using the k-Nearest Neighbor method which will be optimized the model with the k-Fold Cross Validation algorithm. Evaluation of the k-NN model with the k-FCV algorithm can improve accuracy and can build better machine learning models in the image classification process. The default K-NN obtained an accuracy rate of 57%, while the results of the model evaluation with the k-FCV algorithm, on fold 3 obtained an accuracy rate of 68%. The percentage yield with the new model increased by 11% which indicates that the machine learning model that was built was quite optimal