K-Pop Idol Industry 20100663 Minhyung Lee
1. K-Pop Idol History 2. Idol Industry Factor 3. Regression Analysis 4. Result & Interpretation
K-Pop Idol History (1990s) Turning point of Korean Music history : Incorporation of American popular music styles. Firstly introduced by Seotaji & Boys(1992) Starting of the Idol Training system by Lee, Su-man, the founder and producer of the SM Entertainment. - 1st generation of Idol stars came out : H.O.T.(1996) S.E.S.(1997), Sechs Kies(1997), FIN.K.L(1998), Shinhwa(1999), god(1999)
K-Pop Idol History(2000s) Early 2000s : Starting of Korean flow upon Japan & Adaption of systematic idol training syste - BoA(2000) Mid of 2000s : New revival and rising of idol industry - DBSK(2003), SS501(2005), Super Junior(2005), BIG BANG(2006) End of 2000s : Opening of the Idol Golden Age & Change in the concept of Idol = Total Entertainer & Expansion of Korean Wave - Wonder Girls(2007), Girls' Generation(2007), SHINee(2008), 2NE1(2009), f(x)(2009), 4minute(2009)
K-pop Idols' Golden Age - 2010s Ø Golden Age of Idol Industry More than two hundreds of idol groups Expansion of K-Pop to foreign countries. Korean Wave & Globally big Fan culture Considered as one step to be a star Foreign Market is the main part of their profit Idol industry becomes main profit of music industry... Exports by Contry in US Dollar Year Japan China 2008 $11.2 million $1.80 million 2009 $21.6 million $2.36 million 2012 $204 million $3.60 million K-Pop Industry Statistics Year Total Exports in US $ Youtube Views 2008 $16.5 million 2009 $31.3 million 2010 $84.9 million 800 million 2011 $180 million 2.2 billion 2012 $235 million 7.0 billion
Costs for 5-member Idol Group Montly Fixed Cost Cost for Releasing New Album Rent of Dorm $1,000 Composition Cost $12,000 Dance Room $1,000 Recording Master $8,000 Food & Snack $3,000 Photography $3,000 Dance Lesson $2,000 Album Jacket Design $4,000 Vocal Lesson $3,000 Costume Design $20,000 Acting Lesson $2,000 Music Video Recording $50,000 Beauty & Cosmetics, Plastic Surgery $10,000 Promotion $100,000 Director Casting $3,000 Hair & Make-up $5,000 Staff wage $15,000 Dance Creation $6,000 Back Dancer Wage $7,000 Sum $40,000 Etc $215,000 v More than $400,000 to train groups and release one album in 6 months
K-pop Idols' Golden Age - 2010s But also Red Ocean of Idol Industry Too many idol groups these days - More than 200 groups in total Lots of money spent to make a debute of one idol group Year Number of Groups Debuted 2010 19 2011 30 2012 37 2013 14 v From now on, they need to access to this industry systematically, which properties and factors are needed.
1. K-Pop Idol History 2. Idol Industry Factor 3. Regression Analysis 4. Result & Interpretation
K-pop Idols' Revenue 1.Streaming Music File (MP3 File) 2. Album Sales 3. Concert / Musical Performances 4. MV Advertisements 5. Individual Acts (ex. Drama / Movie / Musical) 6. International Events
Factors related to Idol Industry Ø Outer result of Idol groups Sales amount of Music albums Online Download Sales Number of View of Music Videos on Youtube Ø Internal factor of Each Idol Groups Pretty, Handsome face & Skinny body - TV, Movie, CF Good Singing Skills - Basic skill for singer, Musical Act Gorgeous Dancing Skills - Basic skill for k-pop star - Basic skill for k-pop star Number of Members
Data Collection Music sales amount & online download from Gaon Chart, Korean Bill-board Number of Views from Youtube Dance Level from WAWA Dance academy Information about each groups from Naver Appearance and Popularity from Survey
Data Collection Album Sales & Online Download Sales Sales Data Album Sales Amount from Album Chart of Gaon Chart Music Files Salen Amount from Download Chart in Gaon Chart Some Music Files didn't have the data on the Top 100 Download Chart, even though that albums were on the Top 100 Album Chart I divided the download sales amount by 100,000, so that the scores of download amount of all groups can be put between 0 and 5.
Data Collection Youtube Views & Dance Level Youtube Views Got the youtube music video views from the number shown on the bottom of videos. Divided the number of views according to the year when the song came out, so that I can get the yearly view. Dance Level Dance Level dat from the wawa dance academy, which is one of the famous dance group in Korea They divided the level from 1 to 7. 1 is the easiest and 7 is the hardest.
Data Collection Appearance & Popularity Appearance For appearance data, I got the data of each members in one group and divided the sum of the scores by the number of members, so that I could get the average appearance data. Apperance score is from 1 to 5. 5 is the most gorgeous looking Popularity For Popularity data, I got the data of each group from the survey. Popularity score is from 1 to 10. 10 is the most popular group. For equality, I divided the score according to each groups' debute year, because usually the old groups have more popularity.
Data
1. K-Pop Idol History 2. Idol Industry Factor 3. Regression Analysis 4. Result & Interpretation
Regression Model 1(Album Sales Amount) y = -642157.5 + 37588x1 + 6173.9x2 + 490.1x3 + 10560x4+49186.4x5 y : Album Sales Amount x1 : Popularity x2 : Vocal Level x3 : Dance Level x4 : Appearance x5 : Number of Members Estimate Std. Error t-value Pr(> t ) (Intercept) -642157.5 180489.7-3.558 0.00146 Popularity 37588 21261.2 1.768 0.08881 Vocal 6173.9 2017.3 3.06 0.00508 Dance 490.1 9926.1 0.049 0.961 Appearance 10560 36871.8 0.286 0.77684 Members 49186.4 10645.4 4.62 0.0000915
Regression Model 1 y = -642157.5 + 37588x1 + 6173.9x2 + 490.1x3 + 10560x4+49186.4x5 Residual standard error: 68990 on 26 degrees of freedom Multiple R-squared: 0.6373 Adjusted R-squared: 0.5676 F-statistic: 9.137 on 5 and 26 DF p-value: 0.00004108 Since p-value is very low, this model is significant
Regression Analysis 1 Residual - Estimate Graph Normality Graph In the Residual - Estimate graph, it showed some pattern. Therefore this model is not covariant. In the Normality graph, it follwed the normality pattern.
Hypothesis Test for Subset 1 Ø Hypothesis for Subset Model H 0 : β 3 = β 4 = 0 H 1 : not H 0 X 3 : Dance X 4 : Appearance Reduced Model y = -613522 + 36930x1 + 6232x2 + 50089x5 Estimate Std. Error t-value Pr(> t ) (Intercept) -613522 145042-4.23 0.000226 Popularity 36930 19369 1.907 0.066877 Vocal 6232 1937 3.216 0.003266 Members 50089 9544 5.248 1.41E-05 Residual standard error: 66580 on 28 degrees of freedom Multiple R-squared: 0.6362, Adjusted R-squared: 0.5972 F-statistic: 16.32 on 3 and 28 DF, p-value: 2.509e-06
Hypothesis Test Result 1 Ø Hypothesis for Subset Model H 0 : β 3 = β 4 = 0 H 1 : not H 0 X 3 : Dance X 4 : Appearance Ø Result y = -613522 + 36930x1 + 6232x2 + 50089x5 F = {(124140000000 123746958742) / 2} / (123746958742/26) = 0.04129 0.04129 < F 0.95(2, 26) = 3.37, so H 0 is not rejected As a result, Reduced Model is more adequate.
Regression Model 2 (Online Download Sales) y = -0.99212-0.27406x1 + 0.07736x2-0.17606x3-0.6473x4 + 0.19617x5 y : Online Download Sales x1 : Popularity x2 : Vocal Level x3 : Dance Level x4 : Appearance x5 : Number of Members Estimate Std. Error t-value Pr(> t ) (Intercept) -0.99212 3.51666-0.282 0.7801 Popularity -0.27406 0.41425-0.662 0.5141 Vocal 0.07736 0.03931 1.968 0.0598 Dance -0.17606 0.1934-0.91 0.371 Appearance -0.6473 0.71841-0.901 0.3759 Members 0.19617 0.20741 0.946 0.353
Regression Model 2 Ø y = -0.99212-0.27406x1 + 0.07736x2-0.17606x3-0.6473x4 + 0.19617x5 Residual standard error: 1.344 on 26 degrees of freedom Multiple R-squared: 0.2158, Adjusted R-squared: 0.06505 F-statistic: 1.431 on 5 and 26 DF, p-value: 0.2462 Since p-value is low, this model is significant
Regression Analysis 2 Residual - Estimate Graph Normality Graph In the Residual - Estimate graph, it showed some pattern. Therefore this model is not covariant. In the Normality graph, it slightly follwed the normality pattern.
Hypothesis Test for Subset 2 Ø Hypothesis for Subset Model H 0 : β 1 = β 3 = β 4 = β 5 = 0 H 1 : not H 0 X 1 : Popularity X 3 : Dance X 4 : Appearance X 5 : Number of Members Reduced Model y = -2.33337 + 0.0643x2 Estimate Std. Error t-value Pr(> t ) (Intercept) -2.33337 1.58128-1.476 0.1505 Vocal 0.0643 0.02799 2.297 0.0288 Residual standard error: 1.303 on 30 degrees of freedom Multiple R-squared: 0.1496, Adjusted R-squared: 0.1213 F-statistic: 5.278 on 1 and 30 DF, p-value: 0.02875 Since p-value is low, this model is significant
Hypothesis Test Result 2 Ø Hypothesis for Subset Model H 0 : β 1 = β 3 = β 4 = β 5 = 0 H 1 : not H 0 X 1 : Popularity X 3 : Dance X 4 : Appearance X 5 : Number of Members Ø Result y = -2.33337 + 0.0643x2 F = {(50.946 46.977) / 4} / (46.977 / 26) = 0.549173 0.549173 < F 0.95(4, 26) = 2.74, so H 0 is not rejected As a result, Reduced Model is more adequate.
Regression Model 3 (Youtube Video Views) y = -9412046 + 1344092x1 + 370465x2 + 1436490x3-8468098x4 + 3061916x5 y : Youtube Music Video Views x1 : Popularity x2 : Vocal Level x3 : Dance Level x4 : Appearance x5 : Number of Members Estimate Std. Error t-value Pr(> t ) (Intercept) -9412046 21670847-0.434 0.6676 Popularity 1344092 2552772 0.527 0.603 Vocal 370465 242217 1.529 0.1382 Dance 1436490 1191800 1.205 0.2389 Appearance -8468098 4427090-1.913 0.0668 Members 3061916 1278161 2.396 0.0241
Regression Model 3 Ø y = -9412046 + 1344092x1 + 370465x2 + 1436490x3-8468098x4 + 3061916x5 Residual standard error: 8283000 on 26 degrees of freedom Multiple R-squared: 0.4114 Adjusted R-squared: 0.2982 F-statistic: 3.635 on 5 and 26 DF, p-value: 0.0126 Since p-value is low, this model is significant
Regression Analysis 3 Residual - Estimate Graph Normality Graph In the Residual - Estimate graph, it showed some pattern. Therefore this model is not covariant. In the Normality graph, it follwed the normality pattern.
Hypothesis Test for Subset 3 Ø Hypothesis for Subset Model H 0 : β 1 = β 2 = β 3 = 0 H 1 : not H 0 X 1 : Popularity X 2 : Vocal Level X 3 : Dance Level Reduced Model y = 23146580-9347966x4 + 2715356x5 Estimate Std. Error t-value Pr(> t ) (Intercept) 23146580 14218681 1.628 0.11436 Appearance -9347966 4460609-2.096 0.04495 Members 2715356 854798 3.177 0.00352 Residual standard error: 8582000 on 29 degrees of freedom Multiple R-squared: 0.2953, Adjusted R-squared: 0.2467 F-statistic: 6.075 on 2 and 29 DF, p-value: 0.006256 Since p-value is low, this model is significant
Hypothesis Test Result 3 Ø Hypothesis for Subset Model H 0 : β 1 = β 2 = β 3 = 0 H 1 : not H 0 X 1 : Popularity X 2 : Vocal Level X 3 : Dance Level Ø Result y = 23146580-9347966x4 + 2715356x5 F = {(2136000000000000 1783900000000000) / 3} / (1783900000000000 / 26) = 1.710597 1.710597 < F 0.95(3, 26) = 2.98, so H 0 is not rejected As a result, Reduced Model is more adequate.
1. K-Pop Idol History 2. Idol Industry Factor 3. Regression Analysis 4. Result & Interpretation
Result Analysis Reduced Model 1 (Album Sales Amount) y = -613522 + 36930x1 + 6232x2 + 50089x5 Residual standard error: 66580 on 28 degrees of freedom Multiple R-squared: 0.6362, Adjusted R-squared: 0.5972 p-value: 2.509e-06, No covariance & Normality Reduced Model 2 (Online Download Sales) y = -2.33337 + 0.0643x2 Residual standard error: 1.303 on 30 degrees of freedom Multiple R-squared: 0.1496, Adjusted R-squared: 0.1213 p-value: 0.02875 No covariance & Weak Normality Reduced Model (Youtube Music Video View) y = 23146580-9347966x4 + 2715356x5 Residual standard error: 8582000 on 29 degrees of freedom Multiple R-squared: 0.2953, Adjusted R-squared: 0.2467 p-value: 0.006256 No covariance & Normality Most Significant Model 1 has the biggest R-squared and the smallest p-value Therefore, Model 1 is the most significant
Result Analysis Reduced Model 1 (Album Sales Amount) y = -613522 + 36930x1 + 6232x2 + 50089x5 Popularity Vocal Level # of Members More popular groups have more amount of album sales because of the number of fans will be more than non-popular groups If the vocal level is high in the group, then the range of songs can be wide. Therefore, people can have more interest to their songs. If the number of members are many, they can satisfy different tastes of each fans. As a result, more people will buy those groups' albums
Final Analysis and Limit For the idol groups, the great vocal skill can be considered as the most important factors. Number of members should be considered too when the agency plans to make a new idol group. Appearance preference is very different from person by person, so it doesn't have to be considered as much as vocal skills. Dance level will be different from each group's songs, so the dance level of one song can't be representative of the whole groups' dance skill. Also, most of the input data was very subjective, which means those can be different if I get it from different people. Because of this, there was a problem of lack of covariance.
References http://www.gaonchart.co.kr/chart/download.php?f_type=y ear http://www.gaonchart.co.kr/chart/album.php?f_type=year http://lam01.tistory.com/277 http://blog.naver.com/postview.nhn?blogid=bluecheck9&l ogno=70124216973 http://wawadance.com/ http://www.youtube.com/channel/ucseonk9fs_9jmtw9pw ER9yg
Thank You