Data Mining Untuk Klasifikasi Produk Menggunakan Algoritma K-Nearest Neighbor Pada Toko Online

Abstract

The rapid growth of e-commerce in Indonesia has been largely facilitated by the presence of e-marketplaces. The e-marketplace trend in Indonesia continues to develop along with the development of technology and the internet. During its development, e-marketplaces offer more and more products. As a result, buyers need more effort to find the product they want. In order to facilitate the search for these products, a product classification is carried out. This study classifies products in the Shopee emarketplace using the K-Nearest Neighbor algorithm. The product data used comes from web scraping in the categories of cellphones and accessories, Muslim fashion, and home appliances. The stages of the classification system begin with the preprocessing stage, then the term weighting stage uses the TF-IDF method, then cosine similarity to calculate the similarity distance between documents, and then sorting the results of the cosine similarity to retrieve data for the number of k values. Based on testing on 9 product data with three different k values. Obtained an average that shows the lowest accuracy, precision, and recall results when the value of k = 3. The accuracy result is 88.89%, precision is 83.33%, and a recall of 100% is obtained when using the value of k = 5 or k = 7.