ARIMAX and ARX Models with Social Media Information to Predict Unemployment Rate
Kaaen Kwon 1,
Wan-Sup Cho 2, and
Jonghwa Na 3
1. Department of Business Data Convergence, Chungbuk National University, Cheongju, Korea
2. Department of Management Information System, Chungbuk National University, Cheongju, Korea
3. Department of Information & Statistics/Business Data Convergence, Chungbuk National University, Cheongju, Korea
2. Department of Management Information System, Chungbuk National University, Cheongju, Korea
3. Department of Information & Statistics/Business Data Convergence, Chungbuk National University, Cheongju, Korea
Abstract—To keep current with trends in the society, social media has been actively used for understanding issues and moods of real world. In this paper, we suggest the method using the social media to predict the unemployment rate based on natural language processing and statistical modelling. We adopted AutoRegressive Integrated Moving Average with eXogenous variables (ARIMAX) and AutoRegressive with eXogenous variables (ARX) model to predict the unemployment rate and compared our model to a Google Index based model. Our model derived 27.8% and 27.9% improvements in error reduction compared to an existing model in mean absolute error and mean absolute percentage error metrics, respectively.
Index Terms—social media, unemployment rate, prediction, sentiment analysis, Google Index
Cite: Kaaen Kwon, Wan-Sup Cho, and Jonghwa Na, "ARIMAX and ARX Models with Social Media Information to Predict Unemployment Rate," Journal of Advanced Management Science, Vol. 4, No. 5, pp. 401-404, September 2016. doi: 10.12720/joams.4.5.401-404
Index Terms—social media, unemployment rate, prediction, sentiment analysis, Google Index
Cite: Kaaen Kwon, Wan-Sup Cho, and Jonghwa Na, "ARIMAX and ARX Models with Social Media Information to Predict Unemployment Rate," Journal of Advanced Management Science, Vol. 4, No. 5, pp. 401-404, September 2016. doi: 10.12720/joams.4.5.401-404