library(stats)

# This text file is from the "Unfiltered" sheet from the Excel spreadsheet
# Copy the first 7 columns to a text file, including the headers
# Paste the data in a text file at c:/data/groundhog.txt 
# (or just any text file and change the file reference in this first line)

g <- read.delim("c:/data/groundhog.txt",skip=1)

# Means between the two populations? 
# (the times when Phil predicted "Long Winter" vs. times when he predicted "Early Spring")
# 
mean(g$PAMeanFebTemp[g$Groundhog.Prediction=="Long Winter"])
length(g$PAMeanFebTemp[g$Groundhog.Prediction=="Long Winter"])
mean(g$PAMeanFebTemp[g$Groundhog.Prediction=="Early Spring"])
length(g$PAMeanFebTemp[g$Groundhog.Prediction=="Early Spring"])
#
mean(g$PAMeanMarTemp[g$Groundhog.Prediction=="Long Winter"])
length(g$PAMeanMarTemp[g$Groundhog.Prediction=="Long Winter"])
mean(g$PAMeanMarTemp[g$Groundhog.Prediction=="Early Spring"])
length(g$PAMeanMarTemp[g$Groundhog.Prediction=="Early Spring"])
#
mean(g$PAFebPrecip[g$Groundhog.Prediction=="Long Winter"])
length(g$PAFebPrecip[g$Groundhog.Prediction=="Long Winter"])
mean(g$PAFebPrecip[g$Groundhog.Prediction=="Early Spring"])
length(g$PAFebPrecip[g$Groundhog.Prediction=="Early Spring"])
#
mean(g$PAMarPrecip[g$Groundhog.Prediction=="Long Winter"])
length(g$PAMarPrecip[g$Groundhog.Prediction=="Long Winter"])
mean(g$PAMarPrecip[g$Groundhog.Prediction=="Early Spring"])
length(g$PAMarPrecip[g$Groundhog.Prediction=="Early Spring"])

# Mann-Whitney-Wilcoxon Tests
# Are the metrics different between the two populations? (pred=Y vs. pred=N)
wilcox.test(g$PAMeanFebTemp[g$Groundhog.Prediction=="Long Winter"], g$PAMeanFebTemp[g$Groundhog.Prediction=="Early Spring"])
wilcox.test(g$PAMeanMarTemp[g$Groundhog.Prediction=="Long Winter"], g$PAMeanMarTemp[g$Groundhog.Prediction=="Early Spring"])
wilcox.test(g$PAFebPrecip[g$Groundhog.Prediction=="Long Winter"], g$PAFebPrecip[g$Groundhog.Prediction=="Early Spring"])
wilcox.test(g$PAMarPrecip[g$Groundhog.Prediction=="Long Winter"], g$PAMarPrecip[g$Groundhog.Prediction=="Early Spring"])

# Is Phil's prediction based on January's temperature?
wilcox.test(g$PAMeanJanTemp[g$Groundhog.Prediction=="Long Winter"], g$PAMeanJanTemp[g$Groundhog.Prediction=="Early Spring"])

# Better than Phil - can we predict the outcome of February with the temperature of January?
# Linear regression
summary(lm(g$PAMeanFebTemp ~ g$PAMeanJanTemp))