Prediksi Produktivitas Kakao Kabupaten Mamuju Dengan Algoritma K-Nearest Neighbor

Samsari . (Fakultas Teknik, Universitas Sulawesi Barat)
Irfan A. P (Fakultas Teknik, Universitas Sulawesi Barat)
Nuralamsah Zulkarnaim (Fakultas Teknik, Universitas Sulawesi Barat)

In Search (Informatic, Science, Entrepreneur, Applied Art, Research, Humanism)
Vol 19 No 2 (2020): In Search

This study processes data using the Knowedge Discovery in Database (KDD) process flow. The test sample used was cocoa statistical data in Mamuju district from 2016-2019. The KDD process flow itself consists of data selection, preprocessing, transformation, data mining and evaluation. In the data mining process, determining the pattern of the selected data uses the K-Nearest Neighbor (K-NN) method, while the evaluation uses the Root Mean Square Error (RMSE) method. RMSE is used to measure the level of predictive accuracy of the error value obtained. If the resulting error value is low, it means that the prediction is close to its actual value. The results showed that using the K-NN method can be used to predict cocoa productivity, this is evidenced by the RMSE evaluation of the error value obtained in low districts, with an average error of 6,395.