Start Date
10-12-2017 12:00 AM
Description
This article explores the usefulness of sentiment analysis and Google trends data for car sales forecasting. Previous research has demonstrated the use of both techniques for sales forecasting, but current literature is more ambiguous in its results for forecasting the sales of high involvement goods like cars. In this study, about 500,000 social media posts for eleven car models on the Dutch market are analyzed using linear regression models. Furthermore, this study compares these outcomes to the predictive power of Google Trends. The results suggest that social media sentiments have little predictive power towards car sales while Google Trends data and social mention volume show significant results and can be incorporated into an effective prediction model. A prediction model with time lags using decision tree regression is built that can be used by the car industry as an addition to traditional forecasting methods.
Recommended Citation
Wijnhoven, Fons and Plant, Olivia, "Sentiment Analysis and Google Trends Data for Predicting Car Sales" (2017). ICIS 2017 Proceedings. 1.
https://aisel.aisnet.org/icis2017/DataScience/Presentations/1
Sentiment Analysis and Google Trends Data for Predicting Car Sales
This article explores the usefulness of sentiment analysis and Google trends data for car sales forecasting. Previous research has demonstrated the use of both techniques for sales forecasting, but current literature is more ambiguous in its results for forecasting the sales of high involvement goods like cars. In this study, about 500,000 social media posts for eleven car models on the Dutch market are analyzed using linear regression models. Furthermore, this study compares these outcomes to the predictive power of Google Trends. The results suggest that social media sentiments have little predictive power towards car sales while Google Trends data and social mention volume show significant results and can be incorporated into an effective prediction model. A prediction model with time lags using decision tree regression is built that can be used by the car industry as an addition to traditional forecasting methods.