COMPARISON OF EUCLIDEAN DISTANCE, CAMBERRA DISTANCE, AND CHEBYCHEV DISTANCE IN K-MEANS ALGORITHM BASED ON DBI EVALUATION

Abstract

During the COVID-19 pandemic, almost all businesses experienced difficulties. But not all businesses experience difficulties. Cosmetics is a product category that still exists during the pandemic. Many customers buy cosmetics through online sales. Devi Cosmetics is a trading business which is engaged in selling cosmetics. Due to the large number of sales transactions recorded in the neglected database, it is difficult for business managers to find out which cosmetic products are in high demand by customers and make it difficult for business managers to determine the inventory of cosmetic goods correctly. Determination of the incorrect supply of cosmetics resulted in the loss of the store manager, namely many customers who canceled buying cosmetics due to empty supplies. This study uses the K-Means algorithm to classify sales of cosmetic goods. To find out the best grouping results, it is necessary to compare several distance calculation methods. The distance calculation method here uses three methods, namely Euclidean Distance, Camberra Distance, and Chebychev Distance by finding the DBI value of the three methods. The smallest DBI value is the chebychev distance calculation method with a DBI value = 0.254.