Prediksi harga mobil bekas menggunakan algoritma K-Nearest neighbor

Abstract

The used car market in Indonesia has shown significant growth, as people's need for more economically priced vehicles increases. However, inconsistent price variations often make it difficult for consumers to determine the fair value of a used car. This research aims to build a used car price prediction model using the K-Nearest neighbor (KNN) algorithm.

 The dataset used amounts to 1,079 used car data with selected variables that affect the price of cars such as toyota, honda, mitsubishi brands taken from the oto.com website with data scraping techniques using octoparse, with features such as brand, model, kilometer, location, transmission, type, year, engine size and price.

This research follows the CRISP-DM approach which includes the stages of business understanding, data understanding, data preparation, modeling, and evaluation. Model evaluation is performed using three metrics, namely Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE). by finding the best K value based on the division of 80% testing & test datasets. The results show that the KNN algorithm is able. to predict used car prices with a relatively low error rate. This model is expected to help consumers and automotive industry players in estimating used car market prices more accurately and data-based.