Prediction of Car License Auction Price in Guangzhou: Application of the Time-series Decomposition Model, the Holt-Winters Exponential Smoothing Model and the SARIMA Model

Abstract

This paper aims to predict car license auction prices in Guangzhou and provide useful references on the car license auction prices for the potential participants. Nowadays, China is experiencing the development of urbanization and the economy. Conflict occurs between the increasing population and the limitation of urban spatial resources, which boosts the demand for the expansion of traffic infrastructures. As one of the solutions to relieve the congestion, the car license held by Guangzhou is known as a way to reduce the congestion and provides opportunities for the residents to attain car licenses at the level of prices that they are willing to offer. The Time-series Decomposition Model, the Holt-Winters Exponential Smoothing model (the Holt-Winters model), and the Seasonal Autoregressive Integrated Moving Average model (SARIMA) are employed as tools in data prediction based on the time series data of the car license prices in Guangzhou from January 2013 to December 2021, and predictions are made on the prices on a monthly basis in the year 2022. By describing the features of estimated prices and comparing the MAPE (Mean Absolute Percent Error) from the three models, SARIMA shall be taken as a better model than the Holt-Winters model and the time-series decomposition model with a smaller value of MAPE and a more accurate prediction in the trend of price changing.