value, as shown in Table 5. Taking BYD as an
example, the RMSE and MAPE values of the
SARIMA-BP model based on brand value data are
3458.19 and 0.08 respectively, while the RMSE and
MAPE values of the SARIMA-BP model without
brand value data are 7932.81 and 0.17 respectively.
This shows that the brand quantitative influence
model based on Interbrand has a significant effect on
improving prediction accuracy. At the same time, the
MSE and MAPE values of the SARIMA model are
6527.61 and 0.14 respectively, indicating that the
SARIMA-BP model has improved the model fitting
effect and prediction effect by fitting the nonlinear
part of the residual of the SRAIMA model. The
superiority of the SARIMA-BP neural network model
based on Interbrand's quantitative brand influence in
this study is demonstrated (Yang, 2021).
Table 5: Comparison of model errors (taking BYD as an
example).
Model RMSE MAPE
SARIMA-BP (using brand value
data
3458.19 0.08
SARIMA-BP (brand value data not
used)
7932.81 0.17
SARIMA 6527.61 0.14
3.2 Improvement Plan
The shortcomings of this study are mainly reflected in
data collection and model tuning (Marco et al., 2012).
In terms of data collection, the acquisition of
automobile company financial report data is not
direct, and the problem of inaccurate data is common,
which makes it difficult for the improved Interbrand
brand value model to simulate real data and there are
errors in the calculation of brand value data; the
shortcomings of model tuning are mainly reflected in
the degree of adaptation of the BP neural network
model to the data. In addition, further research can be
supplemented in terms of factors such as new energy
policy subsidies mentioned by Liu (2021), and the
model explanatory variables can be added to improve
the model fitting accuracy (HΓΌlsmann et al., 2012).
4 CONCLUSION
In the prediction of automobile sales, automobile
brand influence plays a vital role in sales. This paper
quantifies the brand influence of new energy vehicles
through the improved Interbrand method, and
integrates it into the SARIMA-BP neural network
model. This method is used to model and predict the
sales time series data of new energy vehicles, which
improves the accuracy of new energy vehicle sales
prediction. Compared with the standard SARIMA and
sales prediction models that do not consider brand
influence data, the model proposed in this study based
on Interbrand quantification of brand influence and
the use of SARIMA-BP neural network model
performs well in RMSE and MAPE indicators. It
provides new ideas for the quantification of
automobile brand influence and sales prediction.
REFERENCES
Chen, D., 2011. Chinese automobile demand prediction
based on ARIMA model. In International Conference
on Biomedical Engineering & Informatics. IEEE.
Fan, Q., 2017. Research on automobile sales forecast based
on SARIMA-BP neural network method. China
Business Review, (22), 3.
Gui, S., Sun, W., & Xu, X., 2021. Automobile sales forecast
analysis based on ARIMA and linear regression
combined model. Computer and Digital Engineering.
HΓΌlsmann, M., Borscheid, D., Friedrich, C. M., 2012.
General sales forecast models for automobile markets
and their analysis. Ibai Publishing, 5(5), 65β86.
Li, J., & Wang, F., 2016. Research on the quantitative
evaluation model of brand value in China based on
Interbrand. Science and Management, 18(1), 5.
Li, N., Zhou, X., Zhang, H., & Deng, S., 2021. Research on
the quantitative evaluation of brand value of
geographical indication agricultural products based on
Interbrand: Taking Zhashui wood ear as an example.
Hubei Agricultural Sciences, 60(10), 156β161.
Liu, Y., Wang, M., & Wang, J., 2021. Research on China's
new energy vehicle market forecast. China Automotive
Industry Review, (2016-4), 86β91.
Marco, H., Borscheid, D., Friedrich, C. M., 2012. General
sales forecast models for automobile markets and their
analysis. Ibai Publishing, 5(5), 65β86.
Pi, J., Zhao, Q., & Peng, J., 2016. CPI combined prediction
model based on SARIMA-GMDH. Statistics and
Decision, (17), 4.γ
Wang, J., Chen, L., Gong, Y., Zheng, Y., Xu, J., & Du, Y., et
al., 2021. A method and device for predicting surface
water quality indicators based on graph neural network.
Journal of Zhejiang University: Engineering Edition,
55(4), 7.
Xu, Y., Wang, Z., & Wu, Z., 2021. A stock trend prediction
model based on CNN-BiLSTM multi-feature fusion.
Modern Library and Information Technology, (007),
005.
Yang, D., 2021. New energy vehicle sales forecast based on
improved BP neural network. Automation and
Instrumentation, (011), 000.
Zhang, J., 2011. Application of SARIMA model in
predicting China's CPI. Statistics and Decision, (5), 3.