ICEWS trend

Load all necessary packages up front

library(scales)
library(gridExtra)
library(ggplot2)
library(lubridate)
library(readr)
library(tidyr)
library(dplyr)

Clear the environment and set the location of the dataset

rm(list = ls())

ICEWS_DATASET <- "~/Projects/koala/results/DatasetCollectorPipeline/events"
GED_DATASET <- "~/Datasets/ucdp/ged/ged30.csv"

Define a function to plot the data

make_plot <- function(title, data) {
  data <- data %>%
    mutate(Date = as.Date(Date)) %>%
    group_by(Date) %>%
    summarise(Count = n()) 
  
  ggplot(data, aes(Date, Count)) +
    ggtitle(title) +
    geom_line(color = "steelblue") +
    theme(axis.text.x = element_text(angle = 30),
          axis.title.x = element_blank(), 
          axis.title.y = element_blank()) +
    scale_x_date(date_breaks = "2 years", 
                 date_minor_breaks = "1 year",
                 date_labels = "%Y") +
    scale_y_continuous(name="Num Events", labels = comma)
}

Create the plots and display them in a grid

plot_trends <- function(events) {
  # create variables for year, month and week from the date field
  events <- events %>%
    mutate(Year = year(EventDate),
           Month = month(EventDate),
           Week = floor((EventDate - min(EventDate)) / 7) + 1)

  daily <- make_plot("Daily", mutate(events, Date = EventDate))
  weekly <- make_plot("Weekly", mutate(events, Date = min(events$EventDate) + weeks(Week)))
  monthly <- make_plot("Monthly", mutate(events, Date = ISOdate(Year, Month, 1)))
  yearly <- make_plot("Yearly", mutate(events, Date = ISOdate(Year, 1, 1)))

  grid.arrange(daily, weekly, monthly, yearly, ncol = 2)
}

Define a function to load ICEWS events

load_icews_events <- function(path, pattern = "^events\\.\\d{4}\\.csv$") {
  files <- list.files(path, pattern = pattern, full.names = TRUE)
  bind_rows(lapply(files, function(f) {
    read_csv(f, progress = FALSE)
  }))
}

ICEWS Events

ICEWS_DATASET %>%
  load_icews_events() %>%
  plot_trends()

GED Events

GED_DATASET %>%
  read_csv() %>%
  rename(EventDate = date_start) %>%
  plot_trends()