Data_Transformation_Script_Billboard.R

library(tidytuesdayR)
library(dplyr)
library(tidyr)
rawdata <- tidytuesdayR::tt_load('2021-09-14')

audio <- rawdata$audio_features
billboard <- rawdata$billboard

billboard_singles <- billboard[0,]
unique_songs <- unique(billboard$song_id)
for(i in 1:length(unique_songs)){
  versions <- filter(billboard, song_id == unique_songs[i])
  min_peak_idx <- which.min(versions$peak_position)
  billboard_singles <- rbind(billboard_singles, versions[min_peak_idx,])
}


billboard_filtered <- select(billboard_singles, -c("song", "performer"))
joined <- merge(billboard_filtered, audio, by="song_id")
joined <- joined %>% select(-c("song_id", "url", "instance", 
                               "spotify_track_id", "spotify_track_preview_url", 
                               "week_position", "previous_week_position"))
joined$date <- as.Date(joined$week_id, "%m/%d/%y")
joined  <-  joined %>% select(-c("week_id"))
joined <- joined %>% drop_na()
write.csv(joined, "joined_billboard_audiofeature.csv")
set.seed(406)
n <- nrow(joined)
training_idxs <- sample.int(n, floor(n * .80), FALSE)
training_set <- joined[training_idxs,]
test_set <- joined[-training_idxs,]
write.csv(training_set, "training.csv")
write.csv(test_set, "test.csv")


train_raw <- read.csv("training.csv")
n <- nrow(train_raw)
trainidx <- sample.int(n, floor(n * .75))
testidx <- setdiff(1:n, trainidx)
train <- train_raw[trainidx, ]
validation <- train_raw[-trainidx,]
write.csv(train, "training.csv")
write.csv(validation, "validation.csv")