Long-term And Short-term Forecasts Of New Energy Vehicle Sales

The new energy automobile industry has been regarded as the primary industry of China's automobile industry transformation and environmental pollution control.According to the periodic fluctuation characteristics of the sales volume of new energy vehicle in China,we analyze its changing rules through different time dimensions,and make long-term and short-term forecasts of the sales volume of new energy vehicle,that is,we discuss the trend of sales changes from the quarterly and monthly time periods respectively.While comparing the prediction accuracy between the models,it can also better control the market and infrastructure construction of new energy vehicles.We have collected raw data samples from 2014 to 2019,six years,24 quarters and 72 months,and fit and analyze the quarterly data using an Autoregressive and Moving Average Model,taking into account the periodicity of the original time series.We modified the ARMA Model into a Seasonal Autoregressive and Moving Average Model(SARMA)and studied the trend of monthly sales growth.Here we also put forward two new models to better and more accurately analyze and regulate the domestic new energy vehicle market.In terms of long-term forecasting,we propose a Grey Forecasting Model based on Data Grouping Approach to reflect the quarterly trend of sales volume.We groupe the original data on the basis of the original grey prediction model(GM(1,1))and establishe a GM(1,1)model based on DGA method,which can effectively identify the quarterly fluctuations of linear data series.In terms of short-term forecasting,we use Singular Spectrum Analysis to perform more accurate analysis,fitting and forecasting of monthly sales of new energy vehicles.This method can facilitate us to extract the principal components of the signals representing various influencing factors,split the variation trend of the original time series into the variation trend of each principal component,so as to better predict the following monthly sales volume through the variation trend of each principal component.According to the results,when making quarterly forecast,the prediction results of ARMA model have a large error with the actual data.Although its fitting effect is good,it is not applicable to the estimation of the future trend of quarterly sales of new energy vehicles.We prefer to use the DGA-GM(1,1)for estimation and prediction.For short-term monthly prediction,SSA model predicting results are good,at the same time considering the relevant domestic policy changes in the second half of 2019,the overall sales volume of new energy vehicles has been affected to some extent,we can in the subsequent study of the model was improved,and the influence factors of the more relevant considerations,make the prediction results more close to real data.