Comparative Analysis of Mathematical Models: A Case Study of Used Smartphone Price Depreciation

The price depression of the smartphone after it becomes used is one of the factors that consumers should consider when buying used smartphones. Therefore, accurate prediction of the price of used smartphones becomes a better insight into marketing strategies, and consumer purchasing decisions. The study aims to compare the performance of the Holt-Winters model damped, logistical decay, and exponential decay to predict the price depression of used smartphones. This research method uses historical data from some used smartphone prices as a basis for analysis. Comparisons between the three models were made to assess the performance of each in representing price depression behavior. Then, a statistical analysis was conducted to determine the most optimal model in predicting the price depression of used smartphones, the optimity of a model measured by the accuracy and speed of the computer performing the model execution. The accuracy measure in this study is the mean square error (MSE) value generated by a model with historical data alignment based on the ratio to the price when the smartphone has not been indicated as usable and the execution speed measure is the time it takes a computer to execute the model. The results show that the Holt-Winters damped model provides the most optimal predictive results compared to the exponential and decay models. This is demonstrated by the lower MSE average value twice as low as compared with the other two models, on the other hand, the holt-winters Damped has the weakness of requiring the longest average execution time of the three models with an average performance speed of 2.7 seconds, but the shortfall is not fatal because the predicted time period in this case is monthly so there is no urgency to advance the speed of seconds rather than accuracy.