Analysing the Refurbished Smart Phone Market with Machine Learning

The refurbished smartphone market has recently attracted attention because of its economic and environmental benefits. In particular, rising environmental awareness and the search for cost-effective alternatives have increased demand for refurbished products. However, the dynamics of this market and its pricing practices differ from those of the new-device market. Price formation depends on several product-specific factors, including device condition and model. Yet, analysing this multi-factor structure and producing accurate price estimates remains challenging for consumers, sellers, and remanufacturers. In this context, machine learning can support high-accuracy price prediction. Developing feature-based price prediction models for refurbished smartphones helps to explain price fluctuations and to estimate a device’s value by considering usage and post-refurbishment condition. In this study, both traditional machine learning and deep learning methods are used to improve prediction accuracy. Model performance is evaluated using MSE, MAE, RMSE, and the R² score. The XGB Regressor achieved the best result among the traditional machine learning algorithms, with an R² of 0.9902. Among the deep learning models, LSTM also performed strongly, reaching an R² of 0.9870.