From 576255116c79f02f92833a8e266e2ab2010f8d83 Mon Sep 17 00:00:00 2001
From: aaron <aaron@duckpond.ch>
Date: Mon, 11 Mar 2024 20:03:52 +0100
Subject: [PATCH] fix tree health plot

---
 ironwood_plots.R | 37 +++++++++++++++++--------------------
 1 file changed, 17 insertions(+), 20 deletions(-)

diff --git a/ironwood_plots.R b/ironwood_plots.R
index 3b234cd..dcd532f 100644
--- a/ironwood_plots.R
+++ b/ironwood_plots.R
@@ -15,32 +15,29 @@ df <- read_ods("ironwood_data_cleaned.ods", sheet = 1)
 # asses the tree health
 ##
 
-# create a vector with our condition names
-condition_names <- c("healthy", "light damage", "medium damage", "severe damage", "at point of death")
+# First, let's create a vector of condition names corresponding to the health index numbers
+condition_names <- c("Healthy", "Light damage", "Medium damage", "Severe damage", "At point of death")
 # Define colors for each condition
 condition_colors <- c("green", "yellow", "orange", "red", "black")
-# Now, let's create the histogram
-hist(df$Tree_Health_Index,
-     breaks = 0:5 - 0.5,  # Setting breaks at midpoints between integers
-     main = "Distribution of Tree Health Index",
-     xlab = "Health Index",
-     ylab = "Number of Trees",
-     col = condition_colors,  # Assigning colors based on the health index values
-     border = "black", # Border color of the bars
-     xlim = c(-0.5, 4.5), # Setting x-axis limits to include all health index values
-     ylim = c(0, max(table(df$Tree_Health_Index))*1.1), # Setting y-axis limits slightly above the maximum frequency
-     axes = FALSE)  # Suppressing axes for customization
-# Add axis labels
-axis(1, at = 0:4, labels = condition_names)
-axis(2)
+# Calculate the percentage of trees in each health condition
+percentage <- prop.table(table(df$Tree_Health_Index)) * 100
+# Now, let's create the bar plot
+barplot(percentage,
+        names.arg = condition_names,
+        main = "Distribution of Tree Health Index",
+        xlab = "Health Index",
+        ylab = "Percentage of Trees",
+        ylim = c(0, max(percentage) + 10),
+        col = condition_colors,
+        border = "black")
 # Adding a legend
-legend("topright", legend = condition_names, fill = condition_colors, border.col = "black")
-# Add a title
-title(main = "Distribution of Tree Health Index")
+legend("topright", legend = condition_names, fill = condition_colors)
 # Adding a grid
-grid(nx = NULL, ny = NULL, col = "lightgray", lty = "dotted")
+#grid(nx = NULL, ny = NULL, col = "lightgray", lty = "dotted")
 # Adding a box around the plot
 box()
+# Add labels with the percentage of trees in each bar
+text(x = barplot(percentage, plot = FALSE), y = percentage, labels = paste0(round(percentage, 1), "%"), pos = 3)
 
 ##
 # Perform a shapiro-wilk normality test