Association rules
Mateusz Pliszka
2024-01-10
Aim of the project
The aim of this project is to find common characteristics between music and mental health problems. I am using a survey data set from kaggle. Technique used for this project is association rules.
library(tidyverse)## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr 1.1.3 ✔ readr 2.1.4
## ✔ forcats 1.0.0 ✔ stringr 1.5.0
## ✔ ggplot2 3.4.4 ✔ tibble 3.2.1
## ✔ lubridate 1.9.3 ✔ tidyr 1.3.0
## ✔ purrr 1.0.2
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errorsData overview and manipulation
First we need to load the data. I am also deleting some variables I know will not be useful.
mxmh_survey_results <- read.csv("C:/Users/mateu/Downloads/mxmh_survey_results.csv")
names(mxmh_survey_results)## [1] "Timestamp" "Age"
## [3] "Primary.streaming.service" "Hours.per.day"
## [5] "While.working" "Instrumentalist"
## [7] "Composer" "Fav.genre"
## [9] "Exploratory" "Foreign.languages"
## [11] "BPM" "Frequency..Classical."
## [13] "Frequency..Country." "Frequency..EDM."
## [15] "Frequency..Folk." "Frequency..Gospel."
## [17] "Frequency..Hip.hop." "Frequency..Jazz."
## [19] "Frequency..K.pop." "Frequency..Latin."
## [21] "Frequency..Lofi." "Frequency..Metal."
## [23] "Frequency..Pop." "Frequency..R.B."
## [25] "Frequency..Rap." "Frequency..Rock."
## [27] "Frequency..Video.game.music." "Anxiety"
## [29] "Depression" "Insomnia"
## [31] "OCD" "Music.effects"
## [33] "Permissions"data <- mxmh_survey_results[,c(-1, -33)]Now, I am looking for any NAs in the dataset.
colSums(is.na(data))## Age Primary.streaming.service
## 1 0
## Hours.per.day While.working
## 0 0
## Instrumentalist Composer
## 0 0
## Fav.genre Exploratory
## 0 0
## Foreign.languages BPM
## 0 107
## Frequency..Classical. Frequency..Country.
## 0 0
## Frequency..EDM. Frequency..Folk.
## 0 0
## Frequency..Gospel. Frequency..Hip.hop.
## 0 0
## Frequency..Jazz. Frequency..K.pop.
## 0 0
## Frequency..Latin. Frequency..Lofi.
## 0 0
## Frequency..Metal. Frequency..Pop.
## 0 0
## Frequency..R.B. Frequency..Rap.
## 0 0
## Frequency..Rock. Frequency..Video.game.music.
## 0 0
## Anxiety Depression
## 0 0
## Insomnia OCD
## 0 0
## Music.effects
## 0It turns out there are a lot of NAs in BPM section. Let’s look into it.
summary(data$BPM)## Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
## 0.00e+00 1.00e+02 1.20e+02 1.59e+06 1.44e+02 1.00e+09 107It turns out some people have put some ridiculous values. We will omit this column anyway so I am going to delete only the row with the age missing.
data <- data[!is.na(data$Age),]Now we can check the structure of the data. We can see right now looking at Music.effects variable that some values have nothing in them but aren’t indicated as NA’s. I am going to explore this issue later on.
str(data)## 'data.frame': 735 obs. of 31 variables:
## $ Age : int 18 63 18 61 18 18 18 21 19 18 ...
## $ Primary.streaming.service : chr "Spotify" "Pandora" "Spotify" "YouTube Music" ...
## $ Hours.per.day : num 3 1.5 4 2.5 4 5 3 1 6 1 ...
## $ While.working : chr "Yes" "Yes" "No" "Yes" ...
## $ Instrumentalist : chr "Yes" "No" "No" "No" ...
## $ Composer : chr "Yes" "No" "No" "Yes" ...
## $ Fav.genre : chr "Latin" "Rock" "Video game music" "Jazz" ...
## $ Exploratory : chr "Yes" "Yes" "No" "Yes" ...
## $ Foreign.languages : chr "Yes" "No" "Yes" "Yes" ...
## $ BPM : int 156 119 132 84 107 86 66 95 94 155 ...
## $ Frequency..Classical. : chr "Rarely" "Sometimes" "Never" "Sometimes" ...
## $ Frequency..Country. : chr "Never" "Never" "Never" "Never" ...
## $ Frequency..EDM. : chr "Rarely" "Never" "Very frequently" "Never" ...
## $ Frequency..Folk. : chr "Never" "Rarely" "Never" "Rarely" ...
## $ Frequency..Gospel. : chr "Never" "Sometimes" "Never" "Sometimes" ...
## $ Frequency..Hip.hop. : chr "Sometimes" "Rarely" "Rarely" "Never" ...
## $ Frequency..Jazz. : chr "Never" "Very frequently" "Rarely" "Very frequently" ...
## $ Frequency..K.pop. : chr "Very frequently" "Rarely" "Very frequently" "Sometimes" ...
## $ Frequency..Latin. : chr "Very frequently" "Sometimes" "Never" "Very frequently" ...
## $ Frequency..Lofi. : chr "Rarely" "Rarely" "Sometimes" "Sometimes" ...
## $ Frequency..Metal. : chr "Never" "Never" "Sometimes" "Never" ...
## $ Frequency..Pop. : chr "Very frequently" "Sometimes" "Rarely" "Sometimes" ...
## $ Frequency..R.B. : chr "Sometimes" "Sometimes" "Never" "Sometimes" ...
## $ Frequency..Rap. : chr "Very frequently" "Rarely" "Rarely" "Never" ...
## $ Frequency..Rock. : chr "Never" "Very frequently" "Rarely" "Never" ...
## $ Frequency..Video.game.music.: chr "Sometimes" "Rarely" "Very frequently" "Never" ...
## $ Anxiety : num 3 7 7 9 7 8 4 5 2 2 ...
## $ Depression : num 0 2 7 7 2 8 8 3 0 2 ...
## $ Insomnia : num 1 2 10 3 5 7 6 5 0 5 ...
## $ OCD : num 0 1 2 3 9 7 0 3 0 1 ...
## $ Music.effects : chr "" "" "No effect" "Improve" ...Now I am looking at distribution of Age to think of a way to split it into a low number of levels.
hist(data$Age, labels = T)
This looks OK.
sum(data$Age<18)## [1] 153sum(data$Age>=18 & data$Age<20)## [1] 146sum(data$Age>=20 & data$Age<23)## [1] 131sum(data$Age>=23 & data$Age<=29)## [1] 145sum(data$Age>29)## [1] 160Now let’s explore the previous issue and check unique values for each column.
lapply(data, unique)## $Age
## [1] 18 63 61 21 19 17 16 15 14 20 26 23 37 36 24 22 42 25 32 29 41 31 33 30 43
## [26] 28 34 13 35 27 53 49 60 38 44 40 64 56 59 10 12 54 57 69 72 73 39 58 67 70
## [51] 55 48 71 74 51 65 80 50 68 46 89
##
## $Primary.streaming.service
## [1] "Spotify" "Pandora"
## [3] "YouTube Music" "I do not use a streaming service."
## [5] "Apple Music" "Other streaming service"
## [7] ""
##
## $Hours.per.day
## [1] 3.00 1.50 4.00 2.50 5.00 1.00 6.00 8.00 2.00 12.00 24.00 0.50
## [13] 0.25 0.00 7.00 10.00 9.00 0.70 4.50 20.00 16.00 15.00 14.00 13.00
## [25] 0.10 11.00 18.00
##
## $While.working
## [1] "Yes" "No" ""
##
## $Instrumentalist
## [1] "Yes" "No" ""
##
## $Composer
## [1] "Yes" "No" ""
##
## $Fav.genre
## [1] "Latin" "Rock" "Video game music" "Jazz"
## [5] "R&B" "K pop" "Country" "EDM"
## [9] "Pop" "Hip hop" "Rap" "Classical"
## [13] "Metal" "Folk" "Lofi" "Gospel"
##
## $Exploratory
## [1] "Yes" "No"
##
## $Foreign.languages
## [1] "Yes" "No" ""
##
## $BPM
## [1] 156 119 132 84 107 86 66
## [8] 95 94 155 NA 125 88 148
## [15] 103 120 99 118 79 169 136
## [22] 101 126 183 142 75 89 140
## [29] 161 80 130 147 77 115 157
## [36] 90 128 96 91 55 85 160
## [43] 108 170 174 127 165 113 167
## [50] 110 150 93 109 175 162 189
## [57] 121 200 105 193 149 134 100
## [64] 98 78 117 122 179 166 144
## [71] 61 185 92 194 60 190 87
## [78] 171 114 141 146 164 145 112
## [85] 111 172 180 76 129 81 52
## [92] 97 143 158 104 135 220 138
## [99] 139 178 62 72 0 208 20
## [106] 124 181 106 191 40 173 123
## [113] 4 73 137 133 102 82 116
## [120] 131 204 159 70 210 8 999999999
## [127] 153 192 56 63 68 624 218
## [134] 186 163 168
##
## $Frequency..Classical.
## [1] "Rarely" "Sometimes" "Never" "Very frequently"
##
## $Frequency..Country.
## [1] "Never" "Sometimes" "Very frequently" "Rarely"
##
## $Frequency..EDM.
## [1] "Rarely" "Never" "Very frequently" "Sometimes"
##
## $Frequency..Folk.
## [1] "Never" "Rarely" "Sometimes" "Very frequently"
##
## $Frequency..Gospel.
## [1] "Never" "Sometimes" "Rarely" "Very frequently"
##
## $Frequency..Hip.hop.
## [1] "Sometimes" "Rarely" "Never" "Very frequently"
##
## $Frequency..Jazz.
## [1] "Never" "Very frequently" "Rarely" "Sometimes"
##
## $Frequency..K.pop.
## [1] "Very frequently" "Rarely" "Sometimes" "Never"
##
## $Frequency..Latin.
## [1] "Very frequently" "Sometimes" "Never" "Rarely"
##
## $Frequency..Lofi.
## [1] "Rarely" "Sometimes" "Very frequently" "Never"
##
## $Frequency..Metal.
## [1] "Never" "Sometimes" "Rarely" "Very frequently"
##
## $Frequency..Pop.
## [1] "Very frequently" "Sometimes" "Rarely" "Never"
##
## $Frequency..R.B.
## [1] "Sometimes" "Never" "Very frequently" "Rarely"
##
## $Frequency..Rap.
## [1] "Very frequently" "Rarely" "Never" "Sometimes"
##
## $Frequency..Rock.
## [1] "Never" "Very frequently" "Rarely" "Sometimes"
##
## $Frequency..Video.game.music.
## [1] "Sometimes" "Rarely" "Very frequently" "Never"
##
## $Anxiety
## [1] 3.0 7.0 9.0 8.0 4.0 5.0 2.0 1.0 6.0 10.0 0.0 7.5
##
## $Depression
## [1] 0.0 2.0 7.0 8.0 3.0 1.0 4.0 5.0 9.0 10.0 6.0 3.5
##
## $Insomnia
## [1] 1.0 2.0 10.0 3.0 5.0 7.0 6.0 0.0 4.0 8.0 9.0 3.5
##
## $OCD
## [1] 0.0 1.0 2.0 3.0 9.0 7.0 4.0 6.0 10.0 5.0 8.0 8.5 5.5
##
## $Music.effects
## [1] "" "No effect" "Improve" "Worsen"As we can see we need to delete the values with empty spaces, but without BPM variable as it will not be taken into account anyway. This way we saved about 100 observations. (Now the number dropped from 735 to 718).
data3 <- data[,-10]
data2 <- data3 %>%
mutate_all(~ifelse(. == "", NA, .)) %>%
mutate(Age = as.character(Age))
cleaned_data <- na.omit(data2)Checking the result
lapply(cleaned_data, unique)## $Age
## [1] "18" "61" "21" "19" "17" "16" "15" "14" "20" "26" "23" "37" "36" "24" "22"
## [16] "25" "32" "29" "41" "31" "33" "30" "43" "28" "34" "13" "35" "27" "53" "49"
## [31] "42" "60" "38" "44" "40" "64" "56" "59" "10" "12" "54" "57" "69" "72" "73"
## [46] "39" "63" "58" "67" "70" "55" "48" "71" "74" "51" "65" "80" "50" "68" "46"
## [61] "89"
##
## $Primary.streaming.service
## [1] "Spotify" "YouTube Music"
## [3] "I do not use a streaming service." "Apple Music"
## [5] "Other streaming service" "Pandora"
##
## $Hours.per.day
## [1] 4.00 2.50 5.00 3.00 1.00 6.00 8.00 2.00 12.00 24.00 0.50 0.25
## [13] 7.00 10.00 1.50 9.00 0.70 4.50 20.00 16.00 15.00 0.00 14.00 0.10
## [25] 11.00 18.00
##
## $While.working
## [1] "No" "Yes"
##
## $Instrumentalist
## [1] "No" "Yes"
##
## $Composer
## [1] "No" "Yes"
##
## $Fav.genre
## [1] "Video game music" "Jazz" "R&B" "K pop"
## [5] "Rock" "Country" "EDM" "Pop"
## [9] "Hip hop" "Rap" "Classical" "Metal"
## [13] "Folk" "Lofi" "Gospel" "Latin"
##
## $Exploratory
## [1] "No" "Yes"
##
## $Foreign.languages
## [1] "Yes" "No"
##
## $Frequency..Classical.
## [1] "Never" "Sometimes" "Rarely" "Very frequently"
##
## $Frequency..Country.
## [1] "Never" "Sometimes" "Very frequently" "Rarely"
##
## $Frequency..EDM.
## [1] "Very frequently" "Never" "Rarely" "Sometimes"
##
## $Frequency..Folk.
## [1] "Never" "Rarely" "Sometimes" "Very frequently"
##
## $Frequency..Gospel.
## [1] "Never" "Sometimes" "Rarely" "Very frequently"
##
## $Frequency..Hip.hop.
## [1] "Rarely" "Never" "Very frequently" "Sometimes"
##
## $Frequency..Jazz.
## [1] "Rarely" "Very frequently" "Never" "Sometimes"
##
## $Frequency..K.pop.
## [1] "Very frequently" "Sometimes" "Never" "Rarely"
##
## $Frequency..Latin.
## [1] "Never" "Very frequently" "Sometimes" "Rarely"
##
## $Frequency..Lofi.
## [1] "Sometimes" "Very frequently" "Rarely" "Never"
##
## $Frequency..Metal.
## [1] "Sometimes" "Never" "Rarely" "Very frequently"
##
## $Frequency..Pop.
## [1] "Rarely" "Sometimes" "Very frequently" "Never"
##
## $Frequency..R.B.
## [1] "Never" "Sometimes" "Very frequently" "Rarely"
##
## $Frequency..Rap.
## [1] "Rarely" "Never" "Very frequently" "Sometimes"
##
## $Frequency..Rock.
## [1] "Rarely" "Never" "Very frequently" "Sometimes"
##
## $Frequency..Video.game.music.
## [1] "Very frequently" "Never" "Rarely" "Sometimes"
##
## $Anxiety
## [1] 7.0 9.0 8.0 4.0 5.0 2.0 1.0 6.0 10.0 3.0 0.0 7.5
##
## $Depression
## [1] 7.0 2.0 8.0 3.0 0.0 1.0 4.0 5.0 9.0 10.0 6.0 3.5
##
## $Insomnia
## [1] 10.0 3.0 5.0 7.0 6.0 0.0 4.0 2.0 8.0 1.0 9.0 3.5
##
## $OCD
## [1] 2.0 3.0 9.0 7.0 0.0 1.0 4.0 6.0 10.0 5.0 8.0 8.5 5.5
##
## $Music.effects
## [1] "No effect" "Improve" "Worsen"Creating levels for daily hours spent for listening to music, Anxiety, Depression, OCD and insomnia variables.
par(mfrow = c(2, 2))
hist(data$Anxiety, labels = T)
hist(data$Depression, labels = T)
hist(data$Insomnia, labels = T)
hist(data$OCD, labels = T)
hist(cleaned_data$Hours.per.day, labels = T)
summary(cleaned_data$Hours.per.day)## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.000 2.000 3.000 3.578 5.000 24.000For the hours variable I am going to use the quartiles. As for the mental health variables I am going to use one type of levels for the sake of similarity of results for each variable. Since I want to differentiate between high-low symptoms observations equal or below 6 are going to be classified as low, and those above as high.
cleaned_data$agecat<-ifelse(cleaned_data[,1]<18, "underage", ifelse(cleaned_data[,1]<20, "age: 18-19", ifelse(cleaned_data[,1]<23, "age: 20-22", ifelse(cleaned_data[,1]<29, "age: 23-28", ifelse(cleaned_data[,1]>=29, "age: older than 28", NA)))))
cleaned_data$anxietycat<-ifelse(cleaned_data[,26]<6, "low anxiety",ifelse(cleaned_data[,26]<11, "high anxiety", NA))
cleaned_data$depressioncat<-ifelse(cleaned_data[,27]<6, "low depression", ifelse(cleaned_data[,27]<11, "high depression", NA))
cleaned_data$insomniacat<-ifelse(cleaned_data[,28]<6, "low insomnia", ifelse(cleaned_data[,28]<11, "high insomnia", NA))
cleaned_data$ocdcat<-ifelse(cleaned_data[,29]<6, "low ocd", ifelse(cleaned_data[,29]<11, "high ocd", NA))
cleaned_data$hourscat<-ifelse(cleaned_data[,3]<2, "under 2 hours", ifelse(cleaned_data[,3]<3, "between 2-3 hours", ifelse(cleaned_data[,3]<5, "between 3-5 hours",ifelse(cleaned_data[,3]>=5, "equal or above 5 hours" , NA))))Now I am transforming variables with two categories into booleans.
cleaned_data1 <- cleaned_data
columns_to_transform <- c(4:6, 8:9)
for (col in columns_to_transform) {
cleaned_data1[, col] <- as.logical(cleaned_data1[, col] == "YES")
}Association rules code
Now we are ready to start using the association rules algorithm. First I am choosing the variables I want to use for the algorithm.
data_AR<-cleaned_data1[, c(2, 4:25, 30:36)]
write.csv(data_AR, file="data_AR.csv")Changing the type of variables into character.
data_AR <-as.data.frame(data_AR)
data_AR1 <- data.frame(lapply(data_AR, as.character), stringsAsFactors = FALSE)
str(data_AR1)## 'data.frame': 718 obs. of 30 variables:
## $ Primary.streaming.service : chr "Spotify" "YouTube Music" "Spotify" "Spotify" ...
## $ While.working : chr "FALSE" "FALSE" "FALSE" "FALSE" ...
## $ Instrumentalist : chr "FALSE" "FALSE" "FALSE" "FALSE" ...
## $ Composer : chr "FALSE" "FALSE" "FALSE" "FALSE" ...
## $ Fav.genre : chr "Video game music" "Jazz" "R&B" "Jazz" ...
## $ Exploratory : chr "FALSE" "FALSE" "FALSE" "FALSE" ...
## $ Foreign.languages : chr "FALSE" "FALSE" "FALSE" "FALSE" ...
## $ Frequency..Classical. : chr "Never" "Sometimes" "Never" "Rarely" ...
## $ Frequency..Country. : chr "Never" "Never" "Never" "Sometimes" ...
## $ Frequency..EDM. : chr "Very frequently" "Never" "Rarely" "Never" ...
## $ Frequency..Folk. : chr "Never" "Rarely" "Never" "Never" ...
## $ Frequency..Gospel. : chr "Never" "Sometimes" "Rarely" "Never" ...
## $ Frequency..Hip.hop. : chr "Rarely" "Never" "Very frequently" "Sometimes" ...
## $ Frequency..Jazz. : chr "Rarely" "Very frequently" "Never" "Very frequently" ...
## $ Frequency..K.pop. : chr "Very frequently" "Sometimes" "Very frequently" "Very frequently" ...
## $ Frequency..Latin. : chr "Never" "Very frequently" "Sometimes" "Rarely" ...
## $ Frequency..Lofi. : chr "Sometimes" "Sometimes" "Sometimes" "Very frequently" ...
## $ Frequency..Metal. : chr "Sometimes" "Never" "Never" "Rarely" ...
## $ Frequency..Pop. : chr "Rarely" "Sometimes" "Sometimes" "Very frequently" ...
## $ Frequency..R.B. : chr "Never" "Sometimes" "Very frequently" "Very frequently" ...
## $ Frequency..Rap. : chr "Rarely" "Never" "Very frequently" "Very frequently" ...
## $ Frequency..Rock. : chr "Rarely" "Never" "Never" "Very frequently" ...
## $ Frequency..Video.game.music.: chr "Very frequently" "Never" "Rarely" "Never" ...
## $ Music.effects : chr "No effect" "Improve" "Improve" "Improve" ...
## $ agecat : chr "age: 18-19" "age: older than 28" "age: 18-19" "age: 18-19" ...
## $ anxietycat : chr "high anxiety" "high anxiety" "high anxiety" "high anxiety" ...
## $ depressioncat : chr "high depression" "high depression" "low depression" "high depression" ...
## $ insomniacat : chr "high insomnia" "low insomnia" "low insomnia" "high insomnia" ...
## $ ocdcat : chr "low ocd" "low ocd" "high ocd" "high ocd" ...
## $ hourscat : chr "between 3-5 hours" "between 2-3 hours" "between 3-5 hours" "equal or above 5 hours" ...Reading the data as transactions using arules library.
library(arules)## Warning: package 'arules' was built under R version 4.3.2## Loading required package: Matrix##
## Attaching package: 'Matrix'## The following objects are masked from 'package:tidyr':
##
## expand, pack, unpack##
## Attaching package: 'arules'## The following object is masked from 'package:dplyr':
##
## recode## The following objects are masked from 'package:base':
##
## abbreviate, writetrans1 <- as(data_AR1, "transactions")## Warning: Column(s) 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17,
## 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30 not logical or factor.
## Applying default discretization (see '? discretizeDF').length(trans1)## [1] 718colnames(trans1)## [1] "Primary.streaming.service=Apple Music"
## [2] "Primary.streaming.service=I do not use a streaming service."
## [3] "Primary.streaming.service=Other streaming service"
## [4] "Primary.streaming.service=Pandora"
## [5] "Primary.streaming.service=Spotify"
## [6] "Primary.streaming.service=YouTube Music"
## [7] "While.working=FALSE"
## [8] "Instrumentalist=FALSE"
## [9] "Composer=FALSE"
## [10] "Fav.genre=Classical"
## [11] "Fav.genre=Country"
## [12] "Fav.genre=EDM"
## [13] "Fav.genre=Folk"
## [14] "Fav.genre=Gospel"
## [15] "Fav.genre=Hip hop"
## [16] "Fav.genre=Jazz"
## [17] "Fav.genre=K pop"
## [18] "Fav.genre=Latin"
## [19] "Fav.genre=Lofi"
## [20] "Fav.genre=Metal"
## [21] "Fav.genre=Pop"
## [22] "Fav.genre=R&B"
## [23] "Fav.genre=Rap"
## [24] "Fav.genre=Rock"
## [25] "Fav.genre=Video game music"
## [26] "Exploratory=FALSE"
## [27] "Foreign.languages=FALSE"
## [28] "Frequency..Classical.=Never"
## [29] "Frequency..Classical.=Rarely"
## [30] "Frequency..Classical.=Sometimes"
## [31] "Frequency..Classical.=Very frequently"
## [32] "Frequency..Country.=Never"
## [33] "Frequency..Country.=Rarely"
## [34] "Frequency..Country.=Sometimes"
## [35] "Frequency..Country.=Very frequently"
## [36] "Frequency..EDM.=Never"
## [37] "Frequency..EDM.=Rarely"
## [38] "Frequency..EDM.=Sometimes"
## [39] "Frequency..EDM.=Very frequently"
## [40] "Frequency..Folk.=Never"
## [41] "Frequency..Folk.=Rarely"
## [42] "Frequency..Folk.=Sometimes"
## [43] "Frequency..Folk.=Very frequently"
## [44] "Frequency..Gospel.=Never"
## [45] "Frequency..Gospel.=Rarely"
## [46] "Frequency..Gospel.=Sometimes"
## [47] "Frequency..Gospel.=Very frequently"
## [48] "Frequency..Hip.hop.=Never"
## [49] "Frequency..Hip.hop.=Rarely"
## [50] "Frequency..Hip.hop.=Sometimes"
## [51] "Frequency..Hip.hop.=Very frequently"
## [52] "Frequency..Jazz.=Never"
## [53] "Frequency..Jazz.=Rarely"
## [54] "Frequency..Jazz.=Sometimes"
## [55] "Frequency..Jazz.=Very frequently"
## [56] "Frequency..K.pop.=Never"
## [57] "Frequency..K.pop.=Rarely"
## [58] "Frequency..K.pop.=Sometimes"
## [59] "Frequency..K.pop.=Very frequently"
## [60] "Frequency..Latin.=Never"
## [61] "Frequency..Latin.=Rarely"
## [62] "Frequency..Latin.=Sometimes"
## [63] "Frequency..Latin.=Very frequently"
## [64] "Frequency..Lofi.=Never"
## [65] "Frequency..Lofi.=Rarely"
## [66] "Frequency..Lofi.=Sometimes"
## [67] "Frequency..Lofi.=Very frequently"
## [68] "Frequency..Metal.=Never"
## [69] "Frequency..Metal.=Rarely"
## [70] "Frequency..Metal.=Sometimes"
## [71] "Frequency..Metal.=Very frequently"
## [72] "Frequency..Pop.=Never"
## [73] "Frequency..Pop.=Rarely"
## [74] "Frequency..Pop.=Sometimes"
## [75] "Frequency..Pop.=Very frequently"
## [76] "Frequency..R.B.=Never"
## [77] "Frequency..R.B.=Rarely"
## [78] "Frequency..R.B.=Sometimes"
## [79] "Frequency..R.B.=Very frequently"
## [80] "Frequency..Rap.=Never"
## [81] "Frequency..Rap.=Rarely"
## [82] "Frequency..Rap.=Sometimes"
## [83] "Frequency..Rap.=Very frequently"
## [84] "Frequency..Rock.=Never"
## [85] "Frequency..Rock.=Rarely"
## [86] "Frequency..Rock.=Sometimes"
## [87] "Frequency..Rock.=Very frequently"
## [88] "Frequency..Video.game.music.=Never"
## [89] "Frequency..Video.game.music.=Rarely"
## [90] "Frequency..Video.game.music.=Sometimes"
## [91] "Frequency..Video.game.music.=Very frequently"
## [92] "Music.effects=Improve"
## [93] "Music.effects=No effect"
## [94] "Music.effects=Worsen"
## [95] "agecat=age: 18-19"
## [96] "agecat=age: 20-22"
## [97] "agecat=age: 23-28"
## [98] "agecat=age: older than 28"
## [99] "agecat=underage"
## [100] "anxietycat=high anxiety"
## [101] "anxietycat=low anxiety"
## [102] "depressioncat=high depression"
## [103] "depressioncat=low depression"
## [104] "insomniacat=high insomnia"
## [105] "insomniacat=low insomnia"
## [106] "ocdcat=high ocd"
## [107] "ocdcat=low ocd"
## [108] "hourscat=between 2-3 hours"
## [109] "hourscat=between 3-5 hours"
## [110] "hourscat=equal or above 5 hours"
## [111] "hourscat=under 2 hours"Checking the results
Now we can look at the results of the algorithm.
rules.high_depression<-apriori(data=trans1, parameter=list(supp=0.001, conf=0.08), appearance=list(default="lhs", rhs="depressioncat=high depression"), control=list(verbose=F))
rules.high_depression.byconf<-sort(rules.high_depression, by="confidence", decreasing=TRUE)
inspect(head(rules.high_depression.byconf, n = 10))## lhs rhs support confidence coverage lift count
## [1] {Primary.streaming.service=Apple Music,
## Fav.genre=Latin} => {depressioncat=high depression} 0.001392758 1 0.001392758 2.130564 1
## [2] {Fav.genre=Latin,
## Frequency..K.pop.=Sometimes} => {depressioncat=high depression} 0.001392758 1 0.001392758 2.130564 1
## [3] {Fav.genre=Latin,
## Frequency..Country.=Sometimes} => {depressioncat=high depression} 0.001392758 1 0.001392758 2.130564 1
## [4] {Fav.genre=Latin,
## agecat=age: 18-19} => {depressioncat=high depression} 0.001392758 1 0.001392758 2.130564 1
## [5] {Fav.genre=Latin,
## Frequency..Lofi.=Sometimes} => {depressioncat=high depression} 0.001392758 1 0.001392758 2.130564 1
## [6] {Fav.genre=Latin,
## Frequency..R.B.=Sometimes} => {depressioncat=high depression} 0.001392758 1 0.001392758 2.130564 1
## [7] {Fav.genre=Latin,
## Frequency..Rap.=Rarely} => {depressioncat=high depression} 0.001392758 1 0.001392758 2.130564 1
## [8] {Fav.genre=Latin,
## Frequency..Hip.hop.=Sometimes} => {depressioncat=high depression} 0.001392758 1 0.001392758 2.130564 1
## [9] {Fav.genre=Latin,
## Frequency..Rock.=Sometimes} => {depressioncat=high depression} 0.001392758 1 0.001392758 2.130564 1
## [10] {Fav.genre=Latin,
## Frequency..Video.game.music.=Never} => {depressioncat=high depression} 0.001392758 1 0.001392758 2.130564 1Here we can observe top 10 baskets (by confidence) where high depression category is on the right hand side. In other words, we can see what might cause the high depression in this data set. We can see Lofi and Gosper genres as favorite are popular in this case.
rules.high_ocd<-apriori(data=trans1, parameter=list(supp=0.001, conf=0.08), appearance=list(default="lhs", rhs="ocdcat=high ocd"), control=list(verbose=F))
rules.high_ocd.byconf<-sort(rules.high_ocd, by="confidence", decreasing=TRUE)
inspect(head(rules.high_ocd.byconf, n = 10))## lhs rhs support confidence coverage lift count
## [1] {Fav.genre=Lofi,
## Frequency..Gospel.=Very frequently} => {ocdcat=high ocd} 0.001392758 1 0.001392758 5.698413 1
## [2] {Fav.genre=Lofi,
## Frequency..Latin.=Very frequently} => {ocdcat=high ocd} 0.001392758 1 0.001392758 5.698413 1
## [3] {Fav.genre=Lofi,
## Frequency..Country.=Very frequently} => {ocdcat=high ocd} 0.002785515 1 0.002785515 5.698413 2
## [4] {Fav.genre=Lofi,
## Frequency..K.pop.=Very frequently} => {ocdcat=high ocd} 0.001392758 1 0.001392758 5.698413 1
## [5] {Fav.genre=Lofi,
## Frequency..Folk.=Very frequently} => {ocdcat=high ocd} 0.001392758 1 0.001392758 5.698413 1
## [6] {Fav.genre=Lofi,
## Frequency..Rap.=Very frequently} => {ocdcat=high ocd} 0.001392758 1 0.001392758 5.698413 1
## [7] {Fav.genre=Lofi,
## agecat=age: 20-22} => {ocdcat=high ocd} 0.004178273 1 0.004178273 5.698413 3
## [8] {Fav.genre=Lofi,
## Frequency..Pop.=Rarely} => {ocdcat=high ocd} 0.001392758 1 0.001392758 5.698413 1
## [9] {Fav.genre=Lofi,
## Frequency..Video.game.music.=Never} => {ocdcat=high ocd} 0.001392758 1 0.001392758 5.698413 1
## [10] {Frequency..Gospel.=Very frequently,
## Frequency..Jazz.=Very frequently} => {ocdcat=high ocd} 0.001392758 1 0.001392758 5.698413 1When it comes to high OCD Lofi as favorite genre is in every case of this top 10. We can also obvserve high frequency of multiple different genres.
rules.high_insomnia<-apriori(data=trans1, parameter=list(supp=0.001, conf=0.08), appearance=list(default="lhs", rhs="insomniacat=high insomnia"), control=list(verbose=F))
rules.high_insomnia.byconf<-sort(rules.high_insomnia, by="confidence", decreasing=TRUE)
inspect(head(rules.high_insomnia.byconf, n = 10))## lhs rhs support confidence coverage lift count
## [1] {Fav.genre=Latin,
## Frequency..Gospel.=Sometimes} => {insomniacat=high insomnia} 0.001392758 1 0.001392758 3.191111 1
## [2] {Fav.genre=Latin,
## Frequency..Hip.hop.=Very frequently} => {insomniacat=high insomnia} 0.001392758 1 0.001392758 3.191111 1
## [3] {Fav.genre=Latin,
## Frequency..Rap.=Very frequently} => {insomniacat=high insomnia} 0.001392758 1 0.001392758 3.191111 1
## [4] {Fav.genre=Latin,
## Frequency..Metal.=Sometimes} => {insomniacat=high insomnia} 0.001392758 1 0.001392758 3.191111 1
## [5] {Fav.genre=Latin,
## Frequency..Pop.=Rarely} => {insomniacat=high insomnia} 0.001392758 1 0.001392758 3.191111 1
## [6] {Fav.genre=Latin,
## agecat=underage} => {insomniacat=high insomnia} 0.001392758 1 0.001392758 3.191111 1
## [7] {Fav.genre=Latin,
## Music.effects=No effect} => {insomniacat=high insomnia} 0.001392758 1 0.001392758 3.191111 1
## [8] {Fav.genre=Latin,
## Frequency..K.pop.=Rarely} => {insomniacat=high insomnia} 0.001392758 1 0.001392758 3.191111 1
## [9] {Fav.genre=Latin,
## Frequency..Video.game.music.=Sometimes} => {insomniacat=high insomnia} 0.001392758 1 0.001392758 3.191111 1
## [10] {Fav.genre=Latin,
## Frequency..Lofi.=Rarely} => {insomniacat=high insomnia} 0.001392758 1 0.001392758 3.191111 1This is interesting. While lofi was popular when it came to OCD, Latin music is popular with high insomnia. We should also note that each basket found has count of 1 only.
rules.high_anxiety<-apriori(data=trans1, parameter=list(supp=0.001, conf=0.08), appearance=list(default="lhs", rhs="anxietycat=high anxiety"), control=list(verbose=F))
rules.high_anxiety.byconf<-sort(rules.high_anxiety, by="confidence", decreasing=TRUE)
inspect(head(rules.high_anxiety.byconf, n = 10))## lhs rhs support confidence coverage lift count
## [1] {Primary.streaming.service=Apple Music,
## Fav.genre=Latin} => {anxietycat=high anxiety} 0.001392758 1 0.001392758 1.654378 1
## [2] {Fav.genre=Latin,
## Frequency..K.pop.=Sometimes} => {anxietycat=high anxiety} 0.001392758 1 0.001392758 1.654378 1
## [3] {Fav.genre=Latin,
## Frequency..Country.=Sometimes} => {anxietycat=high anxiety} 0.001392758 1 0.001392758 1.654378 1
## [4] {Fav.genre=Latin,
## agecat=age: 18-19} => {anxietycat=high anxiety} 0.001392758 1 0.001392758 1.654378 1
## [5] {Fav.genre=Latin,
## Frequency..Lofi.=Sometimes} => {anxietycat=high anxiety} 0.001392758 1 0.001392758 1.654378 1
## [6] {Fav.genre=Latin,
## Frequency..R.B.=Sometimes} => {anxietycat=high anxiety} 0.001392758 1 0.001392758 1.654378 1
## [7] {Fav.genre=Latin,
## Frequency..Rap.=Rarely} => {anxietycat=high anxiety} 0.001392758 1 0.001392758 1.654378 1
## [8] {Fav.genre=Latin,
## Frequency..Hip.hop.=Sometimes} => {anxietycat=high anxiety} 0.001392758 1 0.001392758 1.654378 1
## [9] {Fav.genre=Latin,
## Frequency..Rock.=Sometimes} => {anxietycat=high anxiety} 0.001392758 1 0.001392758 1.654378 1
## [10] {Fav.genre=Latin,
## Frequency..Video.game.music.=Never} => {anxietycat=high anxiety} 0.001392758 1 0.001392758 1.654378 1Here, with high anxiety, latin music as favorite is popular as well. Also, count is really low. It might be due to the fact that not many people indicated they feel high level of anxiety or insomnia. 9 Overall from the Unsupervised Learning technique we can conclude that the type of favorite genre might be the the indicator of mental health problems. Other than that, we could observe second or third indicator in the basket, however they varied a lot.