What Makes a Song a Summer Hit?

Team Members

Josephine Johnson

Donasia Washington

Introduction

Spotify was launched in 2008 in Europe and was released to the United States in 2011. The digital service now houses over 100 million songs, 5 million podcasts, and 350,000 audio books. However, Spotify is notorious for their music exploration and curated playlists. Spotify’s has songs ranging from the present day and all the way back to 1904. In this report we will be looking at their curated playlist titled Billboard Summer Hits from the year 1958 up until the year 2017.

Data Preparation

First lets load the necessary packages and our dataset as well.

library(tidyverse)

Warning: package ‘tidyverse’ was built under R version 4.3.3

library(ggplot2)
library(ggrepel)

Warning: package ‘ggrepel’ was built under R version 4.3.2

library(stringr)

spotify_summer_hits <- "https://raw.githubusercontent.com/reisanar/datasets/master/all_billboard_summer_hits.csv"
summer_bill_hits <- read_csv(spotify_summer_hits)

Here is a quick glimpse into our data:

glimpse(summer_bill_hits)

Rows: 600
Columns: 22
$ danceability     <dbl> 0.518, 0.543, 0.541, 0.408, 0.554, 0.679, 0.663, 0.684, 0.645, 0.388, 0.556, 0.703, 0.843, 0.551, 0.455, 0.588, 0.715, 0.468, 0.463, 0.568, 0.558, 0.6…
$ energy           <dbl> 0.0600, 0.3320, 0.6760, 0.3970, 0.1890, 0.2790, 0.6190, 0.5560, 0.9430, 0.4340, 0.4560, 0.7530, 0.1200, 0.9340, 0.3270, 0.4490, 0.5720, 0.2810, 0.6070…
$ key              <chr> "A#", "C", "C", "A", "E", "G", "F#", "B", "C", "G", "C", "A", "E", "G", "G", "F", "F#", "G#", "G#", "E", "A#", "E", "C#", "G", "D", "F", "C", "G", "E"…
$ loudness         <dbl> -14.887, -11.573, -7.988, -12.536, -14.277, -10.386, -5.731, -10.602, -1.526, -11.997, -17.609, -11.783, -17.305, -6.660, -10.113, -8.782, -8.628, -13…
$ mode             <chr> "major", "major", "major", "major", "major", "major", "major", "major", "major", "major", "major", "major", "major", "minor", "major", "major", "minor…
$ speechiness      <dbl> 0.0441, 0.0317, 0.1350, 0.0300, 0.0279, 0.0384, 0.0334, 0.0377, 0.0393, 0.0354, 0.0295, 0.1350, 0.0788, 0.0481, 0.0328, 0.0339, 0.0527, 0.0354, 0.0299…
$ acousticness     <dbl> 0.9870, 0.6690, 0.1880, 0.8730, 0.9150, 0.6450, 0.3360, 0.4680, 0.3850, 0.7890, 0.5150, 0.8030, 0.6110, 0.8220, 0.8300, 0.7230, 0.0809, 0.8320, 0.4820…
$ instrumentalness <dbl> 7.87e-06, 0.00e+00, 8.03e-01, 0.00e+00, 1.37e-05, 0.00e+00, 8.61e-06, 0.00e+00, 0.00e+00, 9.54e-01, 0.00e+00, 0.00e+00, 2.31e-06, 1.43e-01, 0.00e+00, …
$ liveness         <dbl> 0.1610, 0.1340, 0.1230, 0.2800, 0.1320, 0.1180, 0.0622, 0.0664, 0.3700, 0.7280, 0.3310, 0.0997, 0.1240, 0.4190, 0.0997, 0.0989, 0.3380, 0.2050, 0.1170…
$ valence          <dbl> 0.336, 0.795, 0.911, 0.697, 0.214, 0.854, 0.979, 0.867, 0.965, 0.873, 0.848, 0.921, 0.622, 0.960, 0.324, 0.889, 0.859, 0.479, 0.964, 0.947, 0.303, 0.7…
$ tempo            <dbl> 127.870, 154.999, 76.231, 72.615, 136.714, 117.287, 185.165, 142.779, 147.768, 206.313, 105.290, 177.162, 128.532, 87.055, 106.303, 128.929, 129.897, …
$ track_uri        <chr> "006Ndmw2hHxvnLbJsBFnPx", "5ayybTSXNwcarDtxQKqvWX", "4jmFSkpcqLOUN6scGU6BOO", "3c7KT5CN8uYRaK3xThhdYt", "2urRqmAFhjZKo8Z6sEGzEv", "1tKMOJW1eTIMSNtILF0…
$ duration_ms      <dbl> 216373, 153933, 128360, 162773, 165293, 161253, 150600, 138733, 131720, 153533, 144133, 152067, 126800, 133093, 173840, 157733, 137107, 134453, 146933…
$ time_signature   <dbl> 4, 4, 4, 4, 3, 3, 4, 4, 4, 4, 3, 4, 4, 4, 4, 4, 4, 3, 4, 4, 3, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, …
$ key_mode         <chr> "A# major", "C major", "C major", "A major", "E major", "G major", "F# major", "B major", "C major", "G major", "C major", "A major", "E major", "G mi…
$ playlist_name    <chr> "summer_hits_1958", "summer_hits_1958", "summer_hits_1958", "summer_hits_1958", "summer_hits_1958", "summer_hits_1958", "summer_hits_1958", "summer_hi…
$ playlist_img     <chr> "https://mosaic.scdn.co/640/5e8c49f7a8d161c1d6510999bd867b6a91640dae6488d1b4d3b17500498b1e648e8a15a663ee1cc083350f7f3e709bffbd6bf47bea2d4132c145484af0…
$ track_name       <chr> "Nel blu dipinto di blu", "Poor Little Fool", "Patricia", "Little Star", "My True Love", "Just A Dream", "When (Originally Performed By The Kalin Twin…
$ artist_name      <chr> "Domenico Modugno", "Ricky Nelson", "Pérez Prado", "The Elegants", "Jack Scott", "Jimmy Clanton", "Paris Music", "The Everly Brothers", "Bobby Darin",…
$ album_name       <chr> "Tutto Modugno (Mister Volare)", "Ricky Nelson (Expanded Edition / Remastered)", "El Rey Del Mambo", "Little Star: The Best Of The Elegants", "Best Of…
$ album_img        <chr> "https://i.scdn.co/image/5e8c49f7a8d161c1d6510999bd867b6a91640dae", "https://i.scdn.co/image/f0f2c3321ca683bdc121ba039b98c13bbf37d6b2", "https://i.scd…
$ year             <dbl> 1958, 1958, 1958, 1958, 1958, 1958, 1958, 1958, 1958, 1958, 1959, 1959, 1959, 1959, 1959, 1959, 1959, 1959, 1959, 1959, 1960, 1960, 1960, 1960, 1960, …

Main Questions

When we first looked at the data we had a couple of initial questions.

1. What is the average duration of song per year, and how does that compare to the most popular songs of other years?

2. Who was an artist that had multiple hits over the years?

3. Is there a direct correlation between the danceability of a song and its popularity?

4. Does tempo - by extension mood - have an affect on how popular a song gets?

While we explore our dataset we keep these questions in mind.

Data Exploration

summary(summer_bill_hits)

  danceability        energy           key               loudness           mode            speechiness       acousticness       instrumentalness       liveness      
 Min.   :0.2170   Min.   :0.0600   Length:600         Min.   :-23.574   Length:600         Min.   :0.02330   Min.   :0.0000488   Min.   :0.0000000   Min.   :0.02480  
 1st Qu.:0.5457   1st Qu.:0.4768   Class :character   1st Qu.:-10.947   Class :character   1st Qu.:0.03280   1st Qu.:0.0417250   1st Qu.:0.0000000   1st Qu.:0.08595  
 Median :0.6480   Median :0.6405   Mode  :character   Median : -8.072   Mode  :character   Median :0.04140   Median :0.1620000   Median :0.0000032   Median :0.12400  
 Mean   :0.6407   Mean   :0.6221                      Mean   : -8.587                      Mean   :0.06866   Mean   :0.2665156   Mean   :0.0364316   Mean   :0.17979  
 3rd Qu.:0.7402   3rd Qu.:0.7830                      3rd Qu.: -5.862                      3rd Qu.:0.06990   3rd Qu.:0.4472500   3rd Qu.:0.0007132   3rd Qu.:0.22275  
 Max.   :0.9800   Max.   :0.9890                      Max.   : -1.097                      Max.   :0.51700   Max.   :0.9870000   Max.   :0.9540000   Max.   :0.98900  
    valence           tempo         track_uri          duration_ms     time_signature    key_mode         playlist_name      playlist_img        track_name       
 Min.   :0.0695   Min.   : 62.83   Length:600         Min.   :103386   Min.   :3.000   Length:600         Length:600         Length:600         Length:600        
 1st Qu.:0.4790   1st Qu.:100.22   Class :character   1st Qu.:192887   1st Qu.:4.000   Class :character   Class :character   Class :character   Class :character  
 Median :0.6900   Median :120.01   Mode  :character   Median :226927   Median :4.000   Mode  :character   Mode  :character   Mode  :character   Mode  :character  
 Mean   :0.6488   Mean   :120.48                      Mean   :229434   Mean   :3.972                                                                              
 3rd Qu.:0.8482   3rd Qu.:133.84                      3rd Qu.:257854   3rd Qu.:4.000                                                                              
 Max.   :0.9860   Max.   :210.75                      Max.   :557293   Max.   :5.000                                                                              
 artist_name         album_name         album_img              year     
 Length:600         Length:600         Length:600         Min.   :1958  
 Class :character   Class :character   Class :character   1st Qu.:1973  
 Mode  :character   Mode  :character   Mode  :character   Median :1988  
                                                          Mean   :1988  
                                                          3rd Qu.:2002  
                                                          Max.   :2017  

print(summer_bill_hits)

Reordering the dataset so track_name and artist_name are at the front of the dataset:

Songs are identified by their names or artists. Putting these 2 variables at the front of the dataset in at least one version of the dataframe made it a little easier to read both when organizing the data and for outsiders when looking at the data.

summer_bill_hits %>% 
  select("track_name", "artist_name", everything())

Answer to Question 1

What is the average duration of song per year, and how does that compare to the most popular songs of other years?

Table 1

Find the average song length for the years between 1958 & 2017:

summer_bill_hits %>% 
  group_by(playlist_name) %>% 
  summarize(avg_duration = mean(duration_ms, na.rm = TRUE)) %>% 
  mutate(avg_duration_mins = (avg_duration/1000)/60) %>% 
  select(-avg_duration)

# putting it in a dataframe
summer_hits_duration <- summer_bill_hits %>% 
  group_by(playlist_name) %>% 
  summarize(avg_duration = mean(duration_ms, na.rm = TRUE)) %>% 
  mutate(avg_duration_mins = (avg_duration/1000)/60) %>% 
  select(-avg_duration)

print(summer_hits_duration)

Table 2

summer_hits_duration %>% 
  filter(avg_duration_mins < 3)

Majority of the songs are greater than 3 minutes, this is interesting because when analyzing the data there is an obvious gradual increase in song length over the scope of 60 years.

Graph

Graph to display this:

with_playlistName_mod <- summer_hits_duration %>%
  mutate(playlist_year = str_extract(playlist_name, "\\d+"))
print(with_playlistName_mod)

ggplot(data = with_playlistName_mod, aes(x = playlist_year, y = avg_duration_mins)) +
  geom_point(aes(), color = "#1DB954") +
  theme(axis.text.x = element_blank()) + 
  theme(plot.title = element_text(hjust = 0.5)) + 
  labs(title = "Average Duration per Year", 
       x = "Years from 1958 to 2017", 
       y = "Average Duration (mins)")

The average duration of popular songs throughout the years can tell us a lot about the tastes and preferences of listeners. There is a very clear positive to negative correlation when looking at the data points. When looking at a dataset that is dependent on the preferences of any population, it is important to keep any outside factors in mind; this is could be due to preference when the music was listened to, world events happening from the period the data was taken from, or general behaviors of the populace (such as attention span).

Answer to Question 2

Who was an artist that had multiple hits over the years?

Table 1

summer_bill_hits %>% 
  group_by(artist_name) %>% 
  count(sort = TRUE)

The artist named “Rihanna” had the most featured songs. Elton John and Katy Perry were a close second.

Table 2

my_girl_riri <- summer_bill_hits %>% 
  filter(artist_name == "Rihanna") %>% 
  select(track_name, danceability, year)
my_girl_riri

Answer to Question 3

Is there a direct correlation between the danceability of a song and its popularity?

Average

Let’s first find the average danceablity over the years:

summer_bill_hits %>% 
  summarize(ave_dance = mean(danceability))

Danceability

The average danceability seems to be 0.64. The range is from 0.2 to 1.0. This is visualized in the graph below:

ggplot(data = summer_bill_hits,
       aes(x = year, y = danceability)) +
  geom_point() +
  geom_smooth(color = "#1DB954")

Rihanna

Lets take a look at Rihanna’s danceability.

ggplot(my_girl_riri,
         aes(x = year, y = danceability, color = track_name)) +
  geom_point(size = 3, alpha = .7) +
  theme(plot.title = element_text(hjust = 0.5)) + 
    labs(title = "Rihanna's Danceability", 
       x = "Years", 
       y = "Danceability")

The artist Rihanna’s songs are above the danceability average of 0.64. Only two of the seven featured songs have a low danceability score. If you’re familiar with her song Pon De Replay, then you could agree that it is a song that has that “summer” feeling to it.

Top Artists

Let’s check the other frequent artists to see if there is a pattern:

summer_bill_hits %>% 
  filter(artist_name %in% c("Rihanna", "Elton John", "Katy Perry", "Mariah Carey", 
                            "The Rolling Stones", "Usher", "Donna Summer",
                            "The Beatles", "Wings")) %>% 
  ggplot(aes(x = year, y = danceability, color = artist_name)) +
  geom_point(size = 8, alpha = .5) +
  theme(plot.title = element_text(hjust = 0.5)) +
  labs(title = "Top Artist's Danceability", x = "Years", y = "Danceability")

It is clear that the average danceability is a useful factor in determining the popularity of a summer song. The more frequently featured artists have a high average danceability with their songs. This could be a key factor in being chosen for each summer playlist.

Answer to Question 4

Does tempo - by extension mood - have an affect on how popular a song gets?

Table 1

summer_bill_hits %>% 
  select("track_name", "tempo", "energy") %>% 
  group_by(tempo)

Table 2

summer_bill_hits %>% 
  summarize(avg_tempo = mean(tempo, na.rm = TRUE))

my_girl_riri_2 <- summer_bill_hits %>% 
  filter(artist_name == "Rihanna") %>% 
  select(track_name, energy, tempo)

my_girl_riri_2

Graph 1

Energy vs Tempo

ggplot(data = summer_bill_hits,
       aes(x = tempo, y = energy), alpha = .5) +
  geom_point() +
  geom_smooth(method = "lm", se = FALSE) + 
  theme(plot.title = element_text(hjust = 0.5)) +
  labs(title = "Tempo and Energy", 
       x = "Energy", 
       y = "Tempo")

Graph 2

ggplot(my_girl_riri_2,
         aes(x = energy, y = tempo, color = track_name)) +
  geom_point(size = 3, alpha = .7) +
  theme(plot.title = element_text(hjust = 0.5)) + 
    labs(title = "Rihanna's Energy", 
       x = "Energy", 
       y = "Tempo")

Tempo is the measurement of how fast or slow a song is. The assumption was that a faster song has a larger energy. In Tempo & Energy this is shown to be false. While the data is generally positive, there is not a very strong correlation between the two variables. A lot of songs on the list have a slower tempo, but make up for it in energy, as shown in Rihanna's Energy with the song Pon de Replay. While it is not a fast tempoed, it more than makes up for it energy and danceability as shown in Rihanna's Danceability.

Conclusion

Overall, Spotify goes through a lot of data to create their curated playlists. Many songs appear to be under the duration of 3 minutes long in early years, with a spike in the late 1900’s before decreasing again in the 2000’s. This could correlate with any outside factors as mentioned above - one of the most discussed being short attention spans in the more recent years. Also, the danceablity has a large impact on the popularity. The most frequent artists have high level of danceable songs. The more upbeat, short, and danceable a song is, the more likely it is to be featured on a summer Spotify playlist.

---

12/06/2023