Research On Automobile Sales Forecasting Model Of ARIMA-RNN Based On Sentiment Analysis

Since the 21st century,China's automobile industry has developed rapidly,and great achievements have been made in terms of scale,sales volume and structural adjustment.However,in recent years,with the slowdown of the domestic macroeconomic growth,the automobile industry has begun to enter a downturn.Many companies are facing the development dilemma of overcapacity and falling production and sales.Some enterprises even had to withdraw from the auto market due to severe expansion of production scale and unreasonable production planning arrangements.Therefore,it is very important for automobile companies to accurately predict sales.To make a scientific and reasonable sales forecast,the choice of models and methods is the key.Existing sales forecasting research is mainly based on the historical sales of cars to predict the sales of cars,ignoring the accuracy of online reviews on the forecast of sales.In view of this,this article analyzes user reviews on automotive professional forums,and takes the Volkswagen brand as the main research object,builds an ARIMA sales forecast model based on sentiment analysis,and uses a recurrent neural network to optimize the model to achieve different results.Make accurate forecasts of car sales for brands.The main work of this article is as follows:(1)Establish the sentiment dictionary in the automotive field,and calculate the sentiment value of online word of mouth based on the dictionary.This article expands on the basis of the general sentiment dictionary,builds a relatively comprehensive sentiment dictionary in the automotive field,and calculates the online word-of-mouth sentiment value of cars based on this.(2)An ARIMA-RNN automobile sales forecast model based on sentiment analysis is constructed.The article takes Volkswagen's Internet word-of-mouth and sales data as the object,uses a recurrent neural network(RNN)to optimize the ARIMA model,builds an ARIMA-RNN sales forecast model which based on sentiment analysis,then compared with the ARIMA model and RNN,which verified the feasibility and superiority of the model.(3)The ARIMA-RNN model based on sentiment analysis is used to predict car sales of some brands in 2020.Applying the sales forecasting model constructed in this paper,the automobile sales of each brand are forecasted,and the forecast results are analyzed in detail.(4)Provide countermeasures and suggestions for the production and sales of China's auto industry.By analyzing the prediction results,we find that China's auto industry has problems such as slower growth,product performance needs to be optimized,and after-sales service needs to be strengthened.In view of this,this paper proposes countermeasures and suggestions from product performance,brand Culture,production and sales,etc.to help companies adjust their development strategies in a timely manner,and then achieve the orderly development of the automotive industry.