「Hit Song Science」的出現使得熱門歌曲預測的相關研究在近年不斷出現。伴隨著音樂產業的數位化，許多音樂產品提供者如串流平台業者除了可以透過消費者使用行為之資訊來進行相關歌曲推薦，亦可利用歌曲本身特徵變數進行數據分析，輔助制定其行銷相關策略。
音樂產業價值鏈中議價能力最高者為主流音樂廠牌，以環球音樂（ Universal Music）、索尼音樂（Sony Music Entertainment）以及華納音樂（Warner Music）此三大音樂廠牌及掌握了全球八成音樂版權，簽約歌手及創作者的創作過程亦受制於唱片公司之行銷策略而使得作品經常必須取決於創作及商業兩者之間，而更因為唱片公司擁有極大資源使得簽約歌手及創作者之議價能力相對較小；與此同時，獨立音樂創作者在音樂產業鏈上之議價能力更顯微薄。
本研究以趨勢分析、集群分析及預測模型分析近代熱門音樂特徵趨勢並建立預測模型。透過集群分析以輔助熱門歌曲之定義；而透過預測模型準確率之提升發現美國音樂產業具有巨星現象。因此，本研究以量化研究之結果提供音樂產業產品提供者建議以輔助其制定決策，改善不同價值鏈上個角色之議價能力，以創造音樂產業進步之原動力。

The emergence of "Hit Song Science" has led to the emergence of research on hit song predictions in recent years. Along with the digitalization of the music industry, many music product relevant providers, such as streaming platform operators, can use the information of consumer behavior to make song recommendations, and can also use the song's own characteristic variables for data analysis to boost the development of marketing related strategy.

The highest power of bargaining in the music industry value chain is the mainstream music label, with the three major music labels of Universal Music, Sony Music Entertainment and Warner Music, owning the 80% of the world's music copyrights. The process of creating a song is also subject to the marketing strategy of record companies so that the works have to be limited between creation and business. Because the music label have great resources, the bargaining power of the contracted singers and creators is relatively low. At the same time, independent music creators are more meager in bargaining power in the music industry chain.

This study analyzes trend of popular music features and builds prediction models. Through cluster analysis to assist the definition of hit songs, and through the improvement of the accuracy of prediction models, the US music industry has found a superstar phenomenon. Therefore, this study provides recommendations from the music industry product providers to assist them in making decisions and improving the bargaining power of different roles in different value chain’s roles to create the momentum behind the music industry's progress.

第一章 緒論 1
第二章 文獻探討 4
第一節 熱門歌曲科學 4
第二節 巨星現象 5
第三節 美國音樂產業概況 7
第三章 研究方法 12
第一節 資料來源概述 12
第二節 特徵欄位來源及定義 12
第三節 研究架構圖 16
第四節 集群分析理論 20
第五節 預測模型 21
第四章 研究結果 26
第一節 趨勢分析 26
第二節 PCC分群結果(Period, Count and Continuation, PCC) 27
第三節 模型預測結果 28
第四節 巨星現象 36
第五章 結論與建議 47
第一節 結論 47
第二節 建議 47
第六章 參考文獻 50

中文部分
1. 高敬原。2018。
https://www.bnext.com.tw/article/50513/a-new-spotify-initiative-could-spell-trouble-for-record-labels-or-kill-spotify
2. 陈琍。2006。用於客户分類的K-means算法，欽州學院學報 ； 21卷6期 2006 / 12 / 10 ， P36 – 37。
3. 李國誠、李彥賢、何宗燁。2017。大數據之應用平行K-means演算法--建構股市決策分析。
4. 吳明隆。2006。SPSS統計應用學習實務：問卷分析與應用統計。
5. 呂洝都。2011。應用資料探勘於淋巴癌病人存活預測之模式。
6. 林成德、彭国兰，2007。隨機森林在企業信用評估指標體系確定中的應用。
7. 沈哲緯、蕭震洋、辜炳寰、曹鼎志、鄭錦桐、羅文俊，2014。運用隨機森林探討莫拉克颱風災區土石流發生因子關連性。
 


外文部分
1. IFPI Global Music Report 2018
https://www.ifpi.org/downloads/GMR2018.pdf
2. John Storey, (2006). Cultural Theory and Popular Culture
3. Dhanaraj and Logan, (2005). Automatic Prediction Of Hit Songs
4. Yizhao Ni , Raúl Santos-rodríguez , Matt Mcvicar , Tijl De Bie, (2011). Hit Song Science Once Again a Science?
5. J Fan and A.Casey, (2013). Study of Chinese and UK Hit Songs Prediction.
6. Dorien Herremans, David Martens & Kenneth Sörensen, (2014). Dance Hit Song Prediction
7. Jacob Jolij, (2013). "Feel Good Music"
8. François Pachet, Pierre Roy, (2008). Hit Song Science Is Not Yet a Science
9. Myra Interiano, Kamyar Kazemi, Lijia Wang, Jienian Yang, Zhaoxia Yu and Natalia L. Komarova, (2018). Musical trends and predictability of success in contemporary songs in and out of the top charts
10. Matthew J. Salganik, Peter Sheridan Dodds, Duncan J. Watts, (2006). Experimental Study of Inequality andUnpredictability in an ArtificialCultural Market
11. E. Zangerle, M. Pichl, B. Hupfauf, and G. Specht, (2016). “Canmicroblogs predict music charts” an analysis of therelationship between “nowplaying tweets” and “musiccharts” in Proceedings of International Society for Mu-sic Information Retrieval
12. Oxford English Dictionary Online
https://www.lexico.com/en
13. Rosen, S. (1981). The Economics of Superstars. ”American Economic Review.
14. Moshe Adler, (2006). Stardom and Talent. ”American Economic Review.
15. Jason Toynbee, (2000). Making Popular Music: Musicians, Creativity and Institutions
16. Roy Shuker, (2012). Understanding Popular Music Culture.
17. Hamlen, W. (1994). Variety and Superstardom in Popular Music.
18. Marie Connolly, Alan B. Krueger, (2005). Rockonomics: The Economics of Popular Music.
19. Myra Interiano, Kamyar Kazemi, Lijia Wang, Jienian Yang, Zhaoxia Yu and Natalia L. Komarova, (2018). Musical trends and predictability of success in contemporary songs in and out of the top charts
20. David E. A. Giles, (2005). Superstardom in the US popular music industry revisited.
21. Jesse C Bockstedt, Robert J. Kauffman, Frederick J. Riggins, (2006). The move to artist-led on-line music distribution: A theory-based assessment and prospects for structural changes in the digital music market.
22. Pulverer B,(2010a). Evolution. EMBO.
23. Verto Index™
https://www.vertoanalytics.com/verto-index/
24. JRC Annual Report 2017, (2018)
25. Singh, (2018).
https://www.kaggle.com/miteshsingh/hollywood-music-dataset)
26. Spotify Audio Feature
https://developer.spotify.com/documentation/web-api/reference/tracks/get-audio-features/
27. Peter Peduzzi, John P Concato, Erica Kemper, Theodore Holford, Alvan R. Feinstein, (2006). A simulation study of the number of events per variable in logistic regression analysis.
28. Fawcett, T. (2006). An introduction to ROC analysis.
29. S.Gopal Krishna Patro, Kishore Kumarsahu,(2015). Normalization: A Preprocessing Stage.
30. Isabelle Guyon, Andr ́e Elisseeff, (2003). An Introduction to Variable and Feature Selection
31. Bekkerman, El-Yaniv, Tishby, Winter, (2003) ; Caruana andde Sa, (2003) ; Forman, (2003) ; Weston, Elisseeff, Sch ̈olkopf, Tipping (2003)
32. Pablo , Furlanello, Biasioli, Gasperi, (2006). Recursive feature elimination with random forest for PTR-MS analysis of agroindustrial products.
33. McKay and Fujinaga,(2006). Musical genre classification: Is it worth pursuing and how can it be improved?
34. Selwyn Piramuthu, (2004). Evaluating feature selection methods for learning in data mining applications.
35. J J Hopfield, (1982). Neural networks and physical systems with emergent collective computational abilities.
36. Gerhard Arminger, Petra Stein, (1997). Finite Mixtures of Covariance Structure Models.
37. Yan-yan SONG, Ying LU , (2015). Decision tree methods: applications for classification and prediction
38. D. J. Hand, William Ernest Henley, (1997). Statistical Classification Methods in Consumer Credit Scoring: A Review.
39. W H Wolberg and O L Mangasarian, (1990). Multisurface method of pattern separation for medical diagnosis applied to breast cytology.
40. Cramer, Ford, Hall, (1976). Estimation of toxic hazard—A decision tree approach.
41. Kim W, Spear ED, Ng DT, (2005). Yos9p detects and targets misfolded glycoproteins for ER-associated degradation.
42. J.R. QUINLAN, (1985). Induction of Decision Trees.
43. Azuaje ,Witten and Frank, (2005). Data Mining: Practical Machine Learning Tools and Techniques.
44. Jack V.Tu, (1996). Advantages and disadvantages of using artificial neural networks versus logistic regression for predicting medical outcomes.
45. Mehrnaz Heidari Soureshjani and Ali Mohammad Kimiagari, (2012). Calculating the best cut off point using logistic regression and neural network on credit scoring problem- A case study of a commercial bank.
46. Leo Breiman, Adele Cutler(1996). RANDOM FORESTS.
47. Tin Kam Ho, (1995). Random decision forests.
48. Bagging Predictors, (1994). Bagging Predictors.
49. G Guo, H Wang, D Bell, Y Bi, K Greer, 2003. KNN model-based approach in classification.
50. SB Imandoust, M Bolandraftar , 2013. Application of k-nearest neighbor (knn) approach for predicting economic events: Theoretical background.
51. V Bijalwan, V Kumar, P Kumari, J Pascua, 2014. KNN based machine learning approach for text and document mining.
52. Tianqi Chen and Carlos Guestrin, 2016. Xgboost: A scalable tree boosting system.
53. Cortes, Vapnik, 1995. Support-vector networks.
54. K Mahesh, G Constantinescu, P Moin ,2004. A numerical method for large-eddy simulation in complex geometries.
55. Marten, Lucy, The shiring of East Anglia: an alternative hypothesis, (2008)
56. Craven and Shavlik, (1996) . Extracting Thee-Structured Representations of Thained Networks
57. Giles, (2005). Internet encyclopaedias go head to head.