display_variance.r

library(dplyr)
library(ggplot2)





# Path to input and output data
path.rgb.long <- ("./df_rgb_long.csv")
path.hsi.long <- ("./df_hsi_long.csv")
path.rpt      <- ("./rpt_no_public.csv")

# Variables demonstrating highest to n-th highest repeatability will be displayed
# in each combination of (TREATMENT, GROWTH_STAGE)
n <- 5

# List of varieties to be excluded, if any; otherwise, exclude.variety as "c()"
exclude.variety <-c("P1", "P2", "P3", "P4") 










# Load RGB and HSI data in long format
message("Loading data...")
df.rgb <- read.csv(path.rgb.long)
df.hsi <- read.csv(path.hsi.long)
df.rpt <- read.csv(path.rpt)

# Exclude varieties, as needed
if (length(exclude.variety)>0){
    df.rgb <- df.rgb %>% filter(!VARIETY %in% exclude.variety)
    df.hsi <- df.hsi %>% filter(!VARIETY %in% exclude.variety)
}

# Combine RGB and HSI data
df <- rbind(df.rgb, df.hsi)

# Simplify attributes under df.rpt$View
df.rpt$View <- replace(df.rpt$View, 
                       df.rpt$View %in% c("SideAll", "SideAverage"), 
                       "Side")
df$View <- replace(df$View, 
                   df$View %in% c("SideAll", "SideAverage"), 
                   "Side")

# Choose valid repeatability measurements
df.rpt <- df.rpt[!is.na(df.rpt$rpt),]

for (treatment in unique(df.rpt$TREATMENT)){
    for (growth.stage in unique(df.rpt$GROWTH_STAGE)){
        print(paste0("Generating plots for ", 
                     treatment, ", ", growth.stage, " case..."))

        # Subset data by treatment and growth stage
        df.rpt.temp <- df.rpt[which(df.rpt$TREATMENT==treatment & 
                                    df.rpt$GROWTH_STAGE==growth.stage), ]
        df.rpt.temp <- df.rpt.temp[order(df.rpt.temp$rpt, decreasing=T), ]


        for (i in seq(1, n)){
            # Gather measurements from a variable with n-th highest repeatability
            row <- df.rpt.temp[i,]
            df.temp <- df[which(df$TREATMENT==row$TREATMENT & 
                                df$GROWTH_STAGE==row$GROWTH_STAGE & 
                                df$View==row$View & 
                                df$frame_nr==row$frame_nr & 
                                df$variable==row$variable), ]
            # Draw box plot
            p = ggplot(df.temp, aes(x=VARIETY, y=value)) +
                ggtitle(paste0("Repeatability=", round(row$rpt, 3))) +
                geom_boxplot() + 
                ylab(paste("Variable: ",
                           row$TREATMENT, row$GROWTH_STAGE, 
                           row$View, row$frame_nr, row$variable, 
                           sep='_'))
            path.plot <- paste0(paste('./variance', 
                                      row$TREATMENT, row$GROWTH_STAGE,
                                      i,
                                      row$View, row$frame_nr, row$variable, 
                                      sep='_'), '.png')
            # Save plot
            png(path.plot)
            print(p)
            dev.off()
        }
    }
}



# EOF