Perbandingan Algoritma K-Nearest Neighbor Dengan Decision Tree Dalam Memprediksi Penjualan Makanan Hewan Peliharaan Di Petshop Dore Vet Clinic

Abstract

INTISARI

Memprediksi penjualan sangat penting dalam kemajuan sebuah usaha, terutama dalam penjualan barang yang memiliki tanggal kadaluarsa seperti makanan hewan peliharaan. Ada beberapa algoritma yang digunakan untuk menginformasikan prediksi harga penjualan salah satunya algoritma K-Nearest Neighbor dan algoritma Decision Tree. Dengan metode K-nn, dihasilkan kondisi dari 30 data, 6 data diklasifikasikan terlaris sesuai dengan prediksi yang dilakukan dengan metode k-nn, 3 data dari 6 data diprediksi terlaris  ternyata tidak terlaris, (data urutan 1, 2, 6). 24 data diprediksi tidak terlaris ternyata 10 data sebelumnya diklasifikasikan terlaris (data urutan 22, 5, 16, 26, 28, 19, 17, 20, 23, 24). Dengan metode decision tree algoritma C45, diketahui dari 30 data, merek purina terlaris, ada 4 data daripada royal canin (false negative). Hasil tingkat akurasi decision tree algoritma c45, diketahui true terlaris = 17, false tidak terlaris = 4. False terlaris = 16, true tidak terlaris = 13. Akurasi decision tree algoritma c45 = 83%.            

Kata kunci— algortima K-NN, Algortima C45, Data Mining, Pohon Keputusan, Prediksi

ABSTRACT

Predicting sales is very important in the progress of a business, especially in the sale of goods that have an expiration date such as pet food. There are several algorithms used to inform sales price predictions, one of which is the K-Nearest Neighbor algorithm and the Decision Tree algorithm. With the K-nn method, conditions are generated from 30 data, 6 data are classified as best-selling according to the predictions made by the k-nn method, 3 data from 6 data are predicted to be the best-sellers in fact, (data order 1, 2, 6). The 24 data predicted not bestselling turned out to be the 10 previously classified bestsellers (data sequences 22, 5, 16, 26, 28, 19, 17, 20, 23, 24). With the C45 decision tree algorithm method, it is known that from 30 data, the best-selling purina brand, there are 4 data than royal canin (false negative). The result of the accuracy level of the decision tree algorithm is c45, it is known that best-selling true = 17, false not best-selling = 4. Best-selling false = 16, best-selling true not = 13. Accuracy of decision tree algorithm c45 = 83%.

Keywords—K-NN algorithm, C45 algorithm, Data Mining, Decision Tree, Prediction