KLASIFIKASI MUTU BERAS MENGGUNAKAN K-NN BERBASIS BACKWARD ELIMINATION

Abstract

CLASSIFICATION RICE QUALITY USING K-NN BACKWARD-BASED ELIMINATION

Rice is one of the most important agricultural products. And it is a strategic commodity because almost all Indonesian people need it. Because of the importance of the function of rice as a staple food ingredient, the quality of the rice to be consumed must be ensured to be of high quality. Determination of the quality or quality of rice until now has been done by many previous researchers. However, of the several methods that have been selected, the majority use image processing. Rice quality data processing using data mining is still rarely done. In this study, the dataset that will be used comes from the Probolinggo Regency Agriculture Service database which has 1 special attribute and 9 regular attributes. The attributes are: variety, length, shape, color, taste, rice planting technique, season, pests, PH and quality. The method that will be used in this study is the backward elimination-based K-NN method.

From the results of the analysis and computation of several methods, it can be concluded that the backward elimination-based K-NN method used to increase the accuracy of rice classification resulted in several conclusions. Namely the attributes that affect the determination of rice quality are variety, shape, color, technique, season, pests and PH. While the features that are considered to have no effect are taste and length. As for the data as much as 4952, the best accuracy is 83.08% when k = 1, while for the data as much as 2000, then the best accuracy is 87.70% at k = 1, while for the data as much as 1000, the best accuracy is 98.60%, namely when k=1.