Implementasi Algoritma K-Nearest Neighbor untuk Melakukan Klasifikasi Produk dari beberapa E-marketplace

Isi Artikel Utama

Danny Sebastian

Abstrak

E-marketplace has gained popularity with the Indonesian society resulting in the increment of products offered. Consequently, customers require more effort to search for products. In this study, we classified products from several e-marketplaces. The classification was carried out using TF-IDF method for the weighting, cosine similarity to calculate product similarity distance, and k-nearest neighbor algorithm. Based on the first testing result using 150 product data, the k-nearest neighbor method with k=5 successfully classified 146 data with 4 data classified into the wrong class. This k=5 value gives the best result for this case, with an accuracy of 97.33%. The second testing result using 150 mixed brand product data, the k-nearest neighbor method successfully classified 145 data with 5 data classified into the wrong class. The accuracy of the second testing is 96.67%.

Unduhan

Data unduhan belum tersedia.

Rincian Artikel

Cara Mengutip
[1]
D. Sebastian, “Implementasi Algoritma K-Nearest Neighbor untuk Melakukan Klasifikasi Produk dari beberapa E-marketplace”, JuTISI, vol. 5, no. 1, Mei 2019.
Bagian
Articles