This project aims to use the historical data of Shanghai License Plate Auction, and also the GDP and number of cars owned by individuals in Shanghai to predict the future lowest bidding price for the plate.

I applied Linear regression to check the correlation between all the variables and conducted a formula. Also by using time series analysis, I got an ARIMA model.

The prediction model we used here accurately predicted the lowest dealing price in a certain level, and we can see the lowest dealing price keeps increasing. It will fluctuate upward in a small interval. But because of the limitations of this model, its confidence interval will be wider over time. The initial factors which may cause this situation include the size of the sample, the confidence level and the variability in the sample.

The outliers in the historical lowest dealing price are good examples to explain the effects due to the governmental regulations and human’s behaviour.

The ARIMA model and other statistical analysis techniques are used to help us to understand the nature of the lowest dealing price in the Shanghai license plate auction. It also proved that sometimes the mathematical analysis cannot accurately predict some things happened in the real world. When we want to make a good decision, the best way is to look at it from different views and consider more factors which may influence the result.

This is only a small project which you can used to do some data analysis practice.

If you have any questions about it please email me: SapiensSuwan@gmail.com :)