# Load the data
X <- read.csv("../data/Maze_UniversityOfIllinois.csv")
head(X)
colnames(X)

# 3.1 Create the initial-error factor variable low/high relative to *median*
# Test final attempt for low-high initial error rates.
n = dim(X)[1]
err1 = rep(0,dim(X)[1])
err1.med = median(X$T1E)
for (i in 1:n)
    if (X$T1E[i]<err1.med) {
            err1[i] = "low"
        } else {
            err1[i] = "high"
            }
# Turn the "high" "low" into factors.
err1.factor <- factor(err1)

# Statistical support
# First attempt
t.test(X$T1T~err1.factor)
# Last attempt
t.test(X$T15T~err1.factor)

# Plots
# First attempt
plot(X$T1T~err1.factor)
# Last attempt
plot(X$T15T~err1.factor)


# 3.2 Create list of mean, median, range, sd for final solution time
# separately for the two groups and print.
# High initial error
# Compute
high = X$T1T[err1.factor=="high"]
high.mean = mean(high)
high.median = median(high)
high.range = range(high)
high.sd = sd(high)
high.stats = list(mean = high.mean, median = high.median, range = high.range, sd = high.sd)
# Display
print("=============================")
print("High initial-error statistics")
print("=============================")
print(paste("  mean: ",high.stats$mean))
print(paste("median: ",high.stats$median))
print(paste(" range: ",high.stats$range))
print(paste("    sd: ",high.stats$sd))
print("")