mtt_haum/code/02_preprocessing.R

109 lines
4.2 KiB
R

# TODO: This script is obsolete and needs to be updated!
# setwd("C:/Users/nwickelmaier/Nextcloud/Documents/MDS/2023ss/60100_master_thesis/code")
#source("functions.R")
devtools::load_all("../../../../software/mtt")
small <- TRUE
now <- Sys.time()
# Read data ##############################################################
cat("########## Reading in data... ##########", "\n")
if (small) {
dat0 <- read.table("../data/haum/rawdata_logfiles_small.csv", sep = ";",
header = TRUE)
} else {
dat0 <- read.table("../data/haum/rawdata_logfiles.csv", sep = ";",
header = TRUE)
}
dat0$date <- as.POSIXct(dat0$date)
dat0$glossar <- ifelse(dat0$artwork == "glossar", 1, 0)
# Remove irrelevant events
dat <- subset(dat0, !(dat0$event %in% c("Start Application",
"Show Application")))
save(dat, file = paste0("tmp/dat_", ifelse(small, "small_", "full_"),
format(now, "%Y-%m-%d_%H-%M-%S"), ".RData"))
# Add trace variable #####################################################
cat("########## Adding trace variable... ##########", "\n")
dat1 <- add_trace(dat)
save(dat1, file = paste("tmp/dat1", ifelse(small, "small_", "full_"),
format(now, "%Y-%m-%d_%H-%M-%S"), ".RData"))
# Close events
cat("########## Closing events... ##########", "\n")
c1 <- close_events(dat1, "move")
cat("## --> move events closed.", "\n")
c2 <- close_events(dat1, "flipCard")
cat("## --> flipCard events closed.", "\n")
c3 <- close_events(dat1, "openTopic")
cat("## --> openTopic events closed.", "\n")
c4 <- close_events(dat1, "openPopup")
cat("## --> openPopup events closed.", "\n")
dat2 <- rbind(c1, c2, c3, c4)
dat2 <- dat2[order(dat2$date.start, dat2$fileId.start), ]
# Remove durations when event spans more than one log file, since they are
# not interpretable
#dat2[which(dat2$fileId.start != dat2$fileId.stop), "duration"] <- NA
# Remove all events that do not have a `date.start`
dat2 <- dat2[!is.na(dat2$date.start), ]
rownames(dat2) <- NULL
save(dat2, file = paste("tmp/dat2", ifelse(small, "small_", "full_"),
format(now, "%Y-%m-%d_%H-%M-%S"), ".RData"))
# Add case variable ######################################################
cat("########## Adding case and eventId variables... ##########", "\n")
dat3 <- add_case(dat2)
# Add event ID ###########################################################
dat3$eventId <- seq_len(nrow(dat3))
dat3 <- dat3[, c("fileId.start", "fileId.stop", "eventId", "case",
"trace", "glossar", "event", "artwork",
"date.start", "date.stop", "timeMs.start",
"timeMs.stop", "duration", "topicNumber", "popup",
"x.start", "y.start", "x.stop", "y.stop",
"distance", "scale.start", "scale.stop",
"scaleSize", "rotation.start", "rotation.stop",
"rotationDegree")]
save(dat3, file = paste("tmp/dat3", ifelse(small, "small_", "full_"),
format(now, "%Y-%m-%d_%H-%M-%S"), ".RData"))
# Add trace for move events ##############################################
cat("\n########## Adding trace variable for move events... ##########", "\n")
dat4 <- add_trace_moves(dat3)
save(dat4, file = paste("tmp/dat4", ifelse(small, "small_", "full_"),
format(now, "%Y-%m-%d_%H-%M-%S"), ".RData"))
# Add topics: file names and topics ######################################
cat("########## Adding information about topics... ##########", "\n")
artworks <- unique(dat4$artwork)
# remove artworks without XML information
artworks <- artworks[!artworks %in% c("504", "505")]
topics <- extract_topics(artworks, pattern = paste0(artworks, ".xml"),
path = "../data/haum/ContentEyevisit/eyevisit_cards_light/")
dat5 <- add_topic(dat4, topics = topics)
save(dat5, file = paste("tmp/dat5", ifelse(small, "small_", "full_"),
format(now, "%Y-%m-%d_%H-%M-%S"), ".RData"))
# TODO: Replace artwork with informative strings
# Export data ############################################################
cat("########## Exporting data frame with event logs... ##########", "\n")
write.table(dat5, "../data/haum/event_logfiles.csv", sep = ";",
row.names = FALSE)