Riigieksamite varieeruvus

analysis.r

library(dplyr)
library(ggplot2)
library(reshape2)

# Read data
df <- read.csv("postimees.csv") %>%
  mutate(postimehe_rank=seq(1, length(postimehe_skoor)))

# Prepare ranks matrix
num_samples <- 500
ranks <- matrix()
length(ranks) <- nrow(df) * num_samples
dim(ranks) <- c(nrow(df), num_samples)

scores <- matrix()
length(scores) <- nrow(df) * num_samples
dim(scores) <- c(nrow(df), num_samples)


# Calculate ranks for num_samples
for(i in 1:num_samples) {
  weights <- runif(3)
  weights <- 3 * weights / sum(weights)
  scores[, i] <- weights[1] * df$eesti_keel + weights[2] * df$matemaatika + weights[3] * df$inglise_keel
  ranks[, i] <- rank(-scores[, i])
}

# Convert to better format for plotting
df_ranks <- cbind(df, as.data.frame(ranks)) %>%
  select(-eesti_keel, -matemaatika, -inglise_keel) %>%
  melt(id.vars=c("kool", "postimehe_skoor", "postimehe_rank"))

df_scores <- cbind(df, as.data.frame(scores)) %>%
  select(-eesti_keel, -matemaatika, -inglise_keel) %>%
  melt(id.vars=c("kool", "postimehe_skoor", "postimehe_rank"))

# Plot
ggplot(df_ranks, aes(x=postimehe_rank, y=value)) +
  geom_point(alpha=0.03, shape=16) +
  xlab("Järjekorranumber, Postimees") +
  ylab("Järjekorranumber, simulatsioon") +
  theme_bw() +
  theme(panel.grid.minor=element_blank())
ggsave("plot_point.png")

ggplot(df_ranks, aes(x=postimehe_rank, y=value, group=variable)) +
  geom_line(alpha=0.03) +
  xlab("Järjekorranumber, Postimees") +
  ylab("Järjekorranumber, simulatsioon") +
  theme_bw() +
  theme(panel.grid.minor=element_blank())
ggsave("plot_line.png")

ggplot(df_scores, aes(x=postimehe_skoor, y=value)) +
  geom_point(alpha=0.01, shape=16) +
  scale_x_reverse() +
  xlab("Skoor, Postimees") +
  ylab("Skoor, simulatsioon") +
  theme_bw() +
  theme(panel.grid.minor=element_blank())

postimees.csv