學術產出-Theses
Article View/Open
Publication Export
-
題名 利用Google關鍵字與機器學習預測日本汽車銷量
Predicting Japanese Car Sales with Google Trends and Machine Learning作者 莫柔娜
Mariia, Morozova貢獻者 羅光達<br>楊子霆
Lo, Kuang Ta<br>Yang, Tzu Ting
莫柔娜
Morozova Mariia關鍵詞 機器學習
LASSO
Google關鍵字
提高预测
Machine learning
LASSO
Google trends
Improved forecast日期 2018 上傳時間 12-Jul-2018 17:20:15 (UTC+8) 摘要 Computers and the Internet has been significantly changing our lives over the past few decades and bringing both a lot of opportunities and challenges to our lives. Internet, on the 1 hand, possess a lot of free and important information. For example, information about consumers’ moods and preferences that can be extracted from the Web using Google Trends search index data which is undoubtedly precious for market research and forecast. While computers and their computation abilities using machine learning make it feasible to improve to improve task performance, particularly forecasting and planning. The aim of this research is to utilize both tools – Google Trends data and Least Absolute Shrinkage and Selection Operator (LASSO, a machine learning method) in forecasting Japanese car sales. This paper pursues two main goals: to compare the machine learning method performance with conventional and human-created models and to identify if Google Trend data helps to improve forecasting model for Japanese car sales. From the results of this research it can be concluded that machine learning methods definitely have some positive implications for forecasting. LASSO definitely outperform human-judgment. Generally, LASSO models with optimal penalty size are very comparable in their out of sample prediction accuracy to autoregressive models. LASSO with optimal lambda also creates models that include a limited number which is undoubtedly easier to interpret. Google Trends data should be treated with care. It is, in generally, advised to run LASSO-regression when working with Google data as LASSO is able to identify the right lags for the Google search indexes that is of a critical importance due to the fact that different brands might have different characteristics and different consumers. 參考文獻 Baker, S. R., Bloom, N., & Davis, S. J. (2016). Measuring economic policy uncertainty. The Quarterly Journal of Economics, 131(4), 1593-1636. Bortoli, C., & Combes, S. Contribution from Google Trends for forecasting the short-term economic outlook in France: limited avenues. Choi, H., & Varian, H. (2012). Predicting the present with Google Trends. Economic Record, 88 (s1), 2-9. Diebold, F.X. (2017). Forecasting. Pennsylvania: Department of Economics, University of Pennsylvania. Retrieved from: http://www.ssc.upenn.edu/~fdiebold/Textbooks.html Fantazzini, D., & Toktamysova, Z. (2015). Forecasting German car sales using Google data and multivariate models. International Journal of Production Economics, 170, 97-135. Gevelber, L. (2016, March). The Car-Buying Process: One Consumer`s 900+ Digital Interactions. Retrieved from https://www.thinkwithgoogle.com/consumer-insights/consumer-car-buying-process-reveals-auto-marketing-opportunities/ Google Inc. (2018). How Trends data is adjusted. Retrieved from https://support.google.com/trends/answer/4365533?hl=en&ref_topic=6248052 Hand, C., & Judge, G. (2012). Searching for the picture: forecasting UK cinema admissions using Google Trends data. Applied Economics Letters, 19(11), 1051-1055. Hyndman, R. J., & Koehler, A. B. (2006). Another look at measures of forecast accuracy. International journal of forecasting, 22(4), 679-688. Kotler, P., & Keller, K. L. (2012). Marketing Management. Global Edition 14e, London: Pearson Education Limited 2012 Li, J., & Chen, W. (2014). Forecasting macroeconomic time series: LASSO-based approaches and their forecast combinations with dynamic factor models. International Journal of Forecasting, 30(4), 996-1015. MAE and RMSE — Which Metric is Better? (2016, March 23). Retrieved from https://medium.com/human-in-a-machine-world/mae-and-rmse-which-metric-is-better-e60ac3bde13d Muehlen, M. (2017) Improved Sales Forecasting with Consumer Behavior. IMES, National Chengchi University Sagaert, Y. R., Aghezzaf, E. H., Kourentzes, N., & Desmet, B. (2017). Temporal big data for tire industry tactical sales forecasting. Interfaces. Shi, Y., Liu, X., Kok, S. Y., Rajarethinam, J., Liang, S., Yap, G., ... & Lo, A. (2016). Three-month real-time dengue forecast models: an early warning system for outbreak alerts and policy decision support in Singapore. Environmental health perspectives, 124(9), 1369. Small, G., & Wong, R. (2002). The validity of forecasting. In A Paper for Presentation at the Pacific Rim Real Estate Society International Conference, Christchurch, New Zealand (pp. 1-14). Spiegelm B. (2015, February 10). The Google Trends Data Goldmine. Retrieved from https://marketingland.com/google-trend-goldmine-117626 Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society. Series B (Methodological), 267-288. Willmott, C. J., & Matsuura, K. (2005). Advantages of the mean absolute error (MAE) over the root mean square error (RMSE) in assessing average model performance. Climate research, 30(1), 79-82. Yang T.T. (2018). Machine Learning and Casual Inference [PowerPoint slides]. Retrieved from: https://drive.google.com/file/d/1wUfA6RzcwHkOTId7_dA86-PJ67T6-xgI/view 描述 碩士
國立政治大學
應用經濟與社會發展英語碩士學位學程(IMES)
105266011資料來源 http://thesis.lib.nccu.edu.tw/record/#G0105266011 資料類型 thesis dc.contributor.advisor 羅光達<br>楊子霆 zh_TW dc.contributor.advisor Lo, Kuang Ta<br>Yang, Tzu Ting en_US dc.contributor.author (Authors) 莫柔娜 zh_TW dc.contributor.author (Authors) Morozova Mariia en_US dc.creator (作者) 莫柔娜 zh_TW dc.creator (作者) Mariia, Morozova en_US dc.date (日期) 2018 en_US dc.date.accessioned 12-Jul-2018 17:20:15 (UTC+8) - dc.date.available 12-Jul-2018 17:20:15 (UTC+8) - dc.date.issued (上傳時間) 12-Jul-2018 17:20:15 (UTC+8) - dc.identifier (Other Identifiers) G0105266011 en_US dc.identifier.uri (URI) http://nccur.lib.nccu.edu.tw/handle/140.119/118630 - dc.description (描述) 碩士 zh_TW dc.description (描述) 國立政治大學 zh_TW dc.description (描述) 應用經濟與社會發展英語碩士學位學程(IMES) zh_TW dc.description (描述) 105266011 zh_TW dc.description.abstract (摘要) Computers and the Internet has been significantly changing our lives over the past few decades and bringing both a lot of opportunities and challenges to our lives. Internet, on the 1 hand, possess a lot of free and important information. For example, information about consumers’ moods and preferences that can be extracted from the Web using Google Trends search index data which is undoubtedly precious for market research and forecast. While computers and their computation abilities using machine learning make it feasible to improve to improve task performance, particularly forecasting and planning. The aim of this research is to utilize both tools – Google Trends data and Least Absolute Shrinkage and Selection Operator (LASSO, a machine learning method) in forecasting Japanese car sales. This paper pursues two main goals: to compare the machine learning method performance with conventional and human-created models and to identify if Google Trend data helps to improve forecasting model for Japanese car sales. From the results of this research it can be concluded that machine learning methods definitely have some positive implications for forecasting. LASSO definitely outperform human-judgment. Generally, LASSO models with optimal penalty size are very comparable in their out of sample prediction accuracy to autoregressive models. LASSO with optimal lambda also creates models that include a limited number which is undoubtedly easier to interpret. Google Trends data should be treated with care. It is, in generally, advised to run LASSO-regression when working with Google data as LASSO is able to identify the right lags for the Google search indexes that is of a critical importance due to the fact that different brands might have different characteristics and different consumers. en_US dc.description.tableofcontents Table of Contents 1.1 Background 1 1.2 Problem statement 2 1.3 Research goal 3 2. Literature review 5 2.1 Forecasting with Google Trends 5 2.2 Forecasting with Google Trends 6 2.3 Forecasting with LASSO Overview 9 3. Data Collection 11 3.1 Japan new cars monthly sales data 11 3.2 Macroeconomic Indicators Data 13 3.3 Google Trends Data 15 4. Methodology 18 4.1 Human Judgement Model Construction 18 4.2 Machine Learning Model 20 4.3 Model Prediction Accuracy Measurements 21 5. Results 24 5.1 Choosing Model by Human Judgement 24 5.2 Choosing Model by Machine Learning Method 28 5.3 Models Comparison: Machine and Human Models 32 5.4 Further Models Comparison 36 6. Discussion of Results 44 References 47 Appendix 1 49 Appendix 2 50 Appendix 3 51 Appendix 4 52 zh_TW dc.format.extent 1656016 bytes - dc.format.mimetype application/pdf - dc.source.uri (資料來源) http://thesis.lib.nccu.edu.tw/record/#G0105266011 en_US dc.subject (關鍵詞) 機器學習 zh_TW dc.subject (關鍵詞) LASSO zh_TW dc.subject (關鍵詞) Google關鍵字 zh_TW dc.subject (關鍵詞) 提高预测 zh_TW dc.subject (關鍵詞) Machine learning en_US dc.subject (關鍵詞) LASSO en_US dc.subject (關鍵詞) Google trends en_US dc.subject (關鍵詞) Improved forecast en_US dc.title (題名) 利用Google關鍵字與機器學習預測日本汽車銷量 zh_TW dc.title (題名) Predicting Japanese Car Sales with Google Trends and Machine Learning en_US dc.type (資料類型) thesis en_US dc.relation.reference (參考文獻) Baker, S. R., Bloom, N., & Davis, S. J. (2016). Measuring economic policy uncertainty. The Quarterly Journal of Economics, 131(4), 1593-1636. Bortoli, C., & Combes, S. Contribution from Google Trends for forecasting the short-term economic outlook in France: limited avenues. Choi, H., & Varian, H. (2012). Predicting the present with Google Trends. Economic Record, 88 (s1), 2-9. Diebold, F.X. (2017). Forecasting. Pennsylvania: Department of Economics, University of Pennsylvania. Retrieved from: http://www.ssc.upenn.edu/~fdiebold/Textbooks.html Fantazzini, D., & Toktamysova, Z. (2015). Forecasting German car sales using Google data and multivariate models. International Journal of Production Economics, 170, 97-135. Gevelber, L. (2016, March). The Car-Buying Process: One Consumer`s 900+ Digital Interactions. Retrieved from https://www.thinkwithgoogle.com/consumer-insights/consumer-car-buying-process-reveals-auto-marketing-opportunities/ Google Inc. (2018). How Trends data is adjusted. Retrieved from https://support.google.com/trends/answer/4365533?hl=en&ref_topic=6248052 Hand, C., & Judge, G. (2012). Searching for the picture: forecasting UK cinema admissions using Google Trends data. Applied Economics Letters, 19(11), 1051-1055. Hyndman, R. J., & Koehler, A. B. (2006). Another look at measures of forecast accuracy. International journal of forecasting, 22(4), 679-688. Kotler, P., & Keller, K. L. (2012). Marketing Management. Global Edition 14e, London: Pearson Education Limited 2012 Li, J., & Chen, W. (2014). Forecasting macroeconomic time series: LASSO-based approaches and their forecast combinations with dynamic factor models. International Journal of Forecasting, 30(4), 996-1015. MAE and RMSE — Which Metric is Better? (2016, March 23). Retrieved from https://medium.com/human-in-a-machine-world/mae-and-rmse-which-metric-is-better-e60ac3bde13d Muehlen, M. (2017) Improved Sales Forecasting with Consumer Behavior. IMES, National Chengchi University Sagaert, Y. R., Aghezzaf, E. H., Kourentzes, N., & Desmet, B. (2017). Temporal big data for tire industry tactical sales forecasting. Interfaces. Shi, Y., Liu, X., Kok, S. Y., Rajarethinam, J., Liang, S., Yap, G., ... & Lo, A. (2016). Three-month real-time dengue forecast models: an early warning system for outbreak alerts and policy decision support in Singapore. Environmental health perspectives, 124(9), 1369. Small, G., & Wong, R. (2002). The validity of forecasting. In A Paper for Presentation at the Pacific Rim Real Estate Society International Conference, Christchurch, New Zealand (pp. 1-14). Spiegelm B. (2015, February 10). The Google Trends Data Goldmine. Retrieved from https://marketingland.com/google-trend-goldmine-117626 Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society. Series B (Methodological), 267-288. Willmott, C. J., & Matsuura, K. (2005). Advantages of the mean absolute error (MAE) over the root mean square error (RMSE) in assessing average model performance. Climate research, 30(1), 79-82. Yang T.T. (2018). Machine Learning and Casual Inference [PowerPoint slides]. Retrieved from: https://drive.google.com/file/d/1wUfA6RzcwHkOTId7_dA86-PJ67T6-xgI/view zh_TW dc.identifier.doi (DOI) 10.6814/THE.NCCU.IMES.001.2018.F06 -
