CalibrationOfCanadianForecastsInStrategicIntelligence.R

## Packages
library(ggplot2)
library(ggthemes)

## Load data
data <- read.csv("canada-accuracy.csv") ## https://open.canada.ca/data/en/dataset/d5f7f93b-f757-42ef-b468-1d3791d97d94
dim(data)
colnames(data)

## Get levels
levels = unique(data[,7])
levels

calibrationList = list()
for(i in c(1:dim(data)[1])){
  level = data[i,7]
  result = data[i,6]
  resultNumeric = ifelse(toString(result) == "Did occur", 1,0)
  ## print(c(toString(result), resultNumeric))
  calibrationList[[toString(level)]] = c(calibrationList[[toString(level)]], resultNumeric)
}
calibrationList

## Analysis
plotdata = list()
plotdata$levels <-  unique(data[,7])
plotdata$historical_frequency <- as.numeric(sapply(calibrationList, mean))
data_frame = as.data.frame(plotdata)

## Plots
ggplot(data = data_frame, aes(x=levels, y=historical_frequency)) +
  geom_point()+
  theme_tufte() + 
  labs(
    title="Calibration of Forecasts in Canadian Strategic Intelligence",
    subtitle="n = 1,514",
    x="Probability estimate", 
    y="Historical frequency" 
  )+  theme(
    legend.title = element_blank(),
    plot.title = element_text(hjust = 0.5),
    plot.subtitle = element_text(hjust = 0.5), 
    legend.box="vertical",
  ) +
  geom_abline(aes(intercept=0, slope=1), color="blue")

height = 5
width = floor(height*(1+sqrt(5))/2)
ggsave("CanadianIntelligence.png", width=width, height=height) ## See plot_as_variable in line 112