#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
# READ RAW DATA INTO R --> SAVE RDATA SUMMARY FOR RELIABILITY AND OTHER CALCULATIONS
# 09/10/2017
# HANNE HUYGELIER
#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

#============================================================================================
# LIBRARIES
#============================================================================================
library(lme4)
library(plyr)
library(psych)
library(moments)

#============================================================================================
# SET Data Directory
#============================================================================================
DataFolder = "C://Users//u0097836//Documents//EFT_Project//Data Manuscript 2//Study 1//"

#============================================================================================
# FUNCTION TO READ IN DATA
#============================================================================================
# FOR CSV FILES:
ReadCSVFile <- function(folder, sep = ",", pattern = ".csv", strsplitsign = "_"){
  
  files <- list.files(path = folder,
                      pattern = pattern,
                      all.files = F,
                      full.names = F)
  
  dataframe <- data.frame()
  
  for(f in 1:length(files)){
    
    data <- read.csv(paste(folder, files[f], sep = ""), header = TRUE, sep = sep, stringsAsFactors = FALSE)
    data$Subject <- as.integer(unlist(strsplit(files[f], strsplitsign))[1])
    dataframe <- rbind(dataframe, data)
    
  }
  
  return(dataframe) 
}

# FOR ; FILES:
ReadCSVFile2 <- function(folder, sep = ";", pattern = ".csv", strsplitsign = "_"){
  
  files <- list.files(path = folder,
                      pattern = pattern,
                      all.files = F,
                      full.names = F)
  
  dataframe <- data.frame()
  
  for(f in 1:length(files)){
    
    data <- read.csv2(paste(folder, files[f], sep = ""), header = TRUE, sep = sep, stringsAsFactors = FALSE)
    data$Subject <- as.integer(unlist(strsplit(files[f], strsplitsign))[1])
    dataframe <- rbind(dataframe, data)
    
  }
  
  return(dataframe) 
}

#============================================================================================
# SUMMARIZE LEFT DATA
#============================================================================================
#----------------------------------------
# # READ RAW DATA INTO A DATAFRAME:
#----------------------------------------
LEFT <- ReadCSVFile(paste0(DataFolder, "EFT//"), pattern = "PR.csv")

# If participant needs more than 1 response, accuracy is not 1:
LEFT$Accuracy <- ifelse(LEFT$NumberofResponses > 1, 0, 1)

# Descriptives of RT's for accurate trials:
describe(LEFT$FirstResponseRT[LEFT$Accuracy == 1])

# TEST SKEW OF RT's:
agostino.test(LEFT$FirstResponseRT[LEFT$Accuracy == 1], alternative = "two.sided")

#-----------------------------------------------------
# MAKE A SUMMARY OF ONLY THE DIFFICULT LEFT ITEMS
#-----------------------------------------------------
# WHAT IS THE MEAN ACCURACY FOR EACH TEST ITEM?
LEFTItems <- ddply(LEFT, c("ImageNames"), function(df)c(MeanAcc = mean(df$Accuracy),
                                                        MedRT = median(df$FirstResponseRT)))
# CHECK THE ASSOCIATION:
plot(LEFTItems$MedRT, LEFTItems$MeanAcc)
abline(h = 0.95, col = "red")
abline(v = 2000, col = "red")

# MAKE A FILTER FOR THE ITEMS:
LEFTItems$Filter <- ifelse(LEFTItems$MedRT < 2000 & LEFTItems$MeanAcc >= 0.95, 0, 1)
LEFTItems$Filter <- ifelse(LEFTItems$MeanAcc == 0, 0, LEFTItems$Filter)

# Histogram of accuracy for those items:
hist(LEFTItems$MeanAcc[LEFTItems$Filter == 1])
hist(LEFTItems$MedRT[LEFTItems$Filter == 1])

# HOW MANY ITEMS ARE THERE THEN?
length(unique(LEFTItems$ImageNames[LEFTItems$Filter == 1]))

# MERGE THIS INFO WITH LEFT RAW DATA:
LEFT2 <- merge(LEFT, LEFTItems, by = "ImageNames")

# REMOVE UNSELECTED ITEMS:
LEFT2 <- LEFT2[LEFT2$Filter == 1, ]

#-----------------------------------------------------
# SPLIT DATA UP INTO TWO PARTS FOR RELIABILITY CHECKS
#-----------------------------------------------------
# LEFT 1 (N = 64):
LEFT$Split <- ifelse(LEFT$ImageNames < 33, 0, 1)

# LEFT 2 (N = 45):
LEFT2$Split <- ifelse(LEFT2$ImageNames < median(unique(LEFT2$ImageNames)), 0, 1)

#----------------------------------------
# SUMMARIZE LEFT DATA:
#----------------------------------------
# SUMMARIZE DATA FOR RELIABILITY CHECK:
LEFTReliability <- ddply(LEFT, c("Subject", "Split"), function(df)c(LEFTAcc = mean(df$Accuracy),
                                                                    LEFTRT = median(df$FirstResponseRT[df$Accuracy == 1])))

LEFTReliability2 <- ddply(LEFT2, c("Subject", "Split"), function(df)c(LEFTAcc = mean(df$Accuracy),
                                                                      LEFTRT = median(df$FirstResponseRT[df$Accuracy == 1])))

# SUMMARIZE DATA:
LEFTValidity <- ddply(LEFT, c("Subject"), function(df)c(LEFTAcc = mean(df$Accuracy),
                                                        LEFTRT = median(df$FirstResponseRT[df$Accuracy == 1])))

LEFTValidity2 <- ddply(LEFT, c("Subject"), function(df)c(LEFTAcc = mean(df$Accuracy),
                                                         LEFTRT = median(df$FirstResponseRT[df$Accuracy == 1])))

#----------------------------------------
# SAVE SUMMARIES OF DATA:
#----------------------------------------
save(LEFTReliability, file = paste0(DataFolder, "Summaries//LEFTReliabilityStudy1.Rdata"))
save(LEFTReliability2, file = paste0(DataFolder, "Summaries//LEFTReliability2Study1.Rdata"))
save(LEFTValidity, file = paste0(DataFolder, "Summaries//LEFTValidityStudy1.Rdata"))
save(LEFTValidity2, file = paste0(DataFolder, "Summaries//LEFTValidity2Study1.Rdata"))

#----------------------------------------
# CLEAN UP:
#----------------------------------------
remove(LEFTReliability, LEFTReliability2, LEFTValidity, LEFTValidity2, LEFT, LEFT2)

#============================================================================================
# SUMMARIZE COHERENT MOTION DATA
#============================================================================================
# READ IN DATA:
CM <- ReadCSVFile(folder = paste0(DataFolder, "CoherentMotion//"), pattern = "CoherentMotion.csv", strsplitsign = "C")

# READ IN DATA:
CM$Subject <- as.character(CM$Subject)
CM$Subject <- substr(CM$Subject, 1, nchar(CM$Subject) - 1)

CM <- CM[CM$MainPrac == 1, ]
CM$TrialNumber <- rep(seq(from = 1, to = 400, by = 1), length(unique(CM$Subject)) + 1)
CM$Split <- ifelse(CM$TrialNumber < 201, 0, 1)

# SUMMARIZE DATA:
CMReliability <- ddply(CM, c("Subject", "Split"), function(df)c(CMAcc = mean(df$Accuracy)))
CMValidity <- ddply(CM, c("Subject"), function(df)c(CMAcc = mean(df$Accuracy)))

#----------------------------------------
# SAVE SUMMARIES OF DATA:
#----------------------------------------
save(CMReliability, file = paste0(DataFolder, "Summaries//CMReliability.Rdata"))
save(CMValidity, file = paste0(DataFolder, "Summaries//CMValidity.Rdata"))

#----------------------------------------
# CLEAN UP:
#----------------------------------------
remove(CMReliability, CMValidity, CM)

#============================================================================================
# SUMMARIZE NAVON DATA
#============================================================================================
# READ IN DATA:
NavonLocal <- ReadCSVFile2(folder = paste0(DataFolder, "Navon//"), sep = ";", pattern = "selAttLoc.csv", strsplitsign = "_")
NavonGlobal <- ReadCSVFile2(folder = paste0(DataFolder, "Navon//"), sep = ";", pattern = "selAttGlob.csv", strsplitsign = "_")

# REMOVE IMPOSSIBLE RT VALUES:
NavonLocal <- NavonLocal[! NavonLocal$rt_raw < 0, ]
NavonGlobal <- NavonGlobal[! NavonGlobal$rt_raw < 0, ]

# RAW RT Distributions:
describe(NavonGlobal$rt_raw[NavonGlobal$acc_raw == 1])
hist(NavonGlobal$rt_raw[NavonGlobal$acc_raw == 1])
describe(NavonLocal$rt_raw[NavonGlobal$acc_raw == 1])
hist(NavonLocal$rt_raw[NavonGlobal$acc_raw == 1])

# TEST FOR SKEWNESS:
agostino.test(NavonGlobal$rt_raw[NavonGlobal$acc_raw == 1], alternative = "two.sided")
agostino.test(NavonLocal$rt_raw[NavonGlobal$acc_raw == 1], alternative = "two.sided")

# COMBINE DATA FROM TWO CONDITIONS:
Navon <- rbind(NavonLocal, NavonGlobal)

# SPLIT INTO TWO HALVES:
Navon$Split <- ifelse(Navon$order < 50, 0, 1)

# MAKE RELIABILITY SUMMARY:
NavonReliability <- ddply(Navon, c("Subject", "cond", "Split"), function(df)c(NavonAcc = mean(df$acc_mean),
                                                                              NavonRT = median(df$rt_raw[df$acc_mean == 1])))

NavonValidity <- ddply(Navon, c("Subject"), function(df)c(NavonLocalAcc = mean(df$acc_raw[df$cond == "selAttLoc"]),
                                                          NavonLocalRT = median(df$rt_raw[df$cond == "selAttLoc" & df$acc_mean == 1]),
                                                          NavonGlobAcc = mean(df$acc_raw[df$cond == "selAttGlob"]),
                                                          NavonGlobRT = median(df$rt_raw[df$cond == "selAttGlob" & df$acc_mean == 1])))

# SAVE SUMMARIES:
save(NavonReliability, file = paste0(DataFolder, "Summaries//NavonReliability.Rdata"))
save(NavonValidity, file = paste0(DataFolder, "Summaries//NavonValidity.Rdata"))