Price Prediction of Second-Hand Iphones Using Random Forest Regression Based on Unit Conditions

Abstract

This study presents the development of a price prediction model for second-hand Iphones based on unit conditions using the Random Forest Regression algorithm, implemented in a web-based application. A dataset of 542 records was collected from Facebook Marketplace and iPhone trading groups, with variables including Iphone type, storage capacity, warranty status, Face ID, and Truetone. The research employed the CRISP-DM methodology through the stages of business understanding, data understanding, data preparation, modeling, evaluation, and deployment. The model was tested using data splits of 80%–20%, 70%–30%, and 60%–40%, resulting in MAE values of 8.32%–8.42% and RMSE values of 10.64%–10.88%, indicating good and consistent accuracy. The developed system can automatically provide price recommendations based on unit conditions, assisting both sellers and buyers in determining fair market prices.