Estimating The Price of Koi Fish Using The K-Nearest Neighbor Method

Abstract

Ornamental fish is one of the fisheries commodities that have high economic value, in the ornamental fish business there are two types of commodities, namely marine and freshwater ornamental fish. One type of freshwater ornamental fish that is popular at this time is koi fish. This fish, which comes from the Cyprinidae family, comes from Japan, and has very attractive patterns and colors and has high and stable economic value. Broadly speaking, koi fish are estimated into 10 categories namely, Kohaku, Sanke, Showa, Bekko, Utsurimono, Asagi, Shusui, Tancho, Hikari and Koromo. Of the many types and categories of Koi fish that exist, there are difficulties for koi fish enthusiasts who are beginners in knowing the price of Koi fish. By using existing data such as the type of fish, fish length, fish age, and local or imported categories, the price of koi fish can be predicted and is expected to help people make decisions in choosing which koi fish to buy. Based on the results of the prediction experiment using the K-Nearest Neighbor algorithm, an accuracy of 67.99% was obtained and the MSE test was 745271550169.4785, MAE was 486469.26012731483, was 863291.1155395256, and R2 was around 68% with a total dataset of 640 records, and the value of k (closest neighbor) produces the highest accuracy i.e. at k=9