UTILIZING CLUSTERING METHODS FOR CATEGORIZING DELIVERY REQUIREMENTS BASED ON ANALYSIS OF E-COMMERCE

The K-Means algorithm model applied has results that show a new insight, namely the grouping of risk levels in the delivery process into 2 clusters, cluster 1 (C0) is a high risk level consisting of 53 data from the 360 datasets tested and cluster 2 (C1) is a low risk level consisting of 307 data from the 360 datasets tested. Testing using the RapidMiner Studio application can also produce similar insights, namely that each cluster has group members who are divided into 2 clusters, namely C0 with 53 data cluster group members, and C1 with 307 data cluster members. Each cluster has an optimal centroid value, namely 131,717 & 385,075 for C0 and 119,932 & 111,414 for C1, with a Davies-Bouldin Index evaluation value of 0.626.