PERBANDINGAN ALGORITMA CART DAN K-NEAREST NEIGHBOR UNTUK PREDIKSI LUAS LAHAN PANEN TANAMAN PADI DI KABUPATEN KARAWANG

Abstract

Karawang regency is known as one of the nation rice granaries because the are many areas of rice fields, especially rice. But the transfer of function from agricultural land into industrial or recidential area can change the geographical structure of Karawang regency previously filled with agricultural land into industrial and property areas. Data mining is a technique of extracting an information from large data. One of them regression techniques. In predicting something a dataset of a numeric data type usually uses a regression technique. In this study used regression techniques to predict the area of harvested land in Karawang regency by using tools WEKA 3.8.2. The resulting comparison is seen from correlation coefficient, mean absolute error, and root mean squared error. In comparison algorithm used the same scenario is cross validation 10 folds. The result of the experiment using the same scenario shows that both algorithm can be used to predict the area of harvest area in Karawang regency. The result of evalution with same scenario shows that CART algorithm has better performance than KNN algorithm with correlation coefficient 0,9646, MAE 498,6229, and RMSE 834,0204