<![CDATA[K-Nearest Neighbor is a classification method that classifies new data into specific classes based on the proximity of characteristics to k members of existing classes. K-Nearest Neighbor relies heavily on training data. In actual circumstances such as the ecommerce customer spending rate dataset, there is no class label for each data. So that to be able to obtain datatraining required additional methods need to be added before the prediction process can be done. This research attempts to use K-Means Clustering to group datasets into multiple clusters which then each cluster will be given a class label according to the centroid characteristics of those clusters. The combination of KNN and K-Means Clustering methods in customer's spending rate predictions gives a fairly good result, where the accuracy of the prediction obtained is 89.6%.]]>
Copyrights © 2021
