R Notebook

github <- "https://raw.githubusercontent.com/RobWHickman/Databases-and-Files/"
  ministers_github <- "master/Great%20Offices/Ministers.csv"
  
Ministers<- read.csv(paste0(github, ministers_github), stringsAsFactors = FALSE)
  Ministers$Startdate <- as.Date(Ministers$Startdate)

library(ggplot2)
library(ggthemes)
library(RColorBrewer)
ggplot(Ministers, aes(Office)) +
  geom_bar(aes(fill = Power), position = "dodge") +
  scale_fill_brewer(palette = "Set2") +
  theme_fivethirtyeight() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
  ggtitle("The Great Offices of State",
          "Number of Holders of Each Office since 1964")

ggplot(Ministers, aes(Office)) +
  geom_bar(aes(fill = paste(Party, Power)), position = "dodge") +
  scale_fill_manual(values = brewer.pal(10, "Set3")[c(5,3,4,6)],
                    guide = guide_legend(title = "Party/Power", ncol = 2)) +
  theme_fivethirtyeight() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
  ggtitle("The Great Offices of State",
          "Number of Holders of Each Office since 1964")

library(stringr)
Ministers <- Ministers[order(Ministers$Startdate),]
duration_func <- function(party, power, office){
    term <- paste (party, power, office)
    which <- which(paste(Ministers$Party, Ministers$Power, Ministers$Office) == term)
      if(grepl("Labour shadow", term)) which <- which[1:(length(which)- 1)]
      if(grepl("Conservative government", term)) which <- which[1:(length(which)- 1)]
    av_days <- mean(Ministers$Duration[which])
    std <- var(Ministers$Duration[which])^0.5
    return(c(Term = term,
             Average_Days = av_days,
             St_Dev = std))
}
durations <- unique(mapply(duration_func, Ministers$Party,
                              Ministers$Power,
                              Ministers$Office,
                    SIMPLIFY = F))
Durations <- do.call(rbind.data.frame,
                     c(durations, stringsAsFactors = FALSE))
  names(Durations) <- c("Holder", "Duration", "Deviation")
  Durations$Duration <- as.numeric(Durations$Duration)
  Durations$Deviation <- as.numeric(Durations$Deviation)
  Durations <- cbind(Durations, data.frame(
                do.call(rbind, c(str_split(Durations$Holder, " ")))))        
  head(Durations)

limits <- aes(ymax = Durations$Duration + Durations$Deviation,
              ymin = Durations$Duration - Durations$Deviation)
ggplot(Durations, aes(Holder, Duration)) +
  geom_bar(aes(fill = paste(X1, X2), alpha = X3), stat = "identity", position = "dodge") +
  scale_fill_manual(values = brewer.pal(10, "Set3")[c(5,3,4,6)],
                    guide = guide_legend(title = "Party/Power", ncol = 2)) +
  scale_x_discrete(labels = rep(unique(Durations$X3), 4)) +
  scale_alpha_manual(values = c("Home" = 0.7, "Foreign" = 0.6, 
                                "Chancellor" = 0.7, "Leader" = 1), guide = "none") +
  geom_errorbar(limits, position= "dodge", width=0.25) +
  theme_fivethirtyeight() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
  ggtitle("The Great Offices of State",
          "Duration of Holders of Each Office since 1964 +/- std.")

library(ggrepel)
elections_github <- "master/Great%20Offices/Elections.csv"
Elections <- read.csv(paste0(github, elections_github), stringsAsFactors = FALSE)
  Elections$Date <- as.Date(Elections$Date)
rectangles <- data.frame(xmin = Elections$Date,
  xmax = append(Elections$Date[2:length(Elections$Date)], as.Date("2017-01-01")),
  ymin = rep(0, length(Elections$Date)),
  ymax = rep(max(Ministers$Duration) + 500, length(Elections$Date)),
  Party <- Elections$Party)
rectangles$Cols <- gsub("Labour", "red", gsub("Conservative", "blue", rectangles$Party))
long <- Ministers[which(Ministers$Duration > 1825),]
ggplot(long) +
  geom_rect(data=rectangles, aes(xmin=xmin, xmax=xmax, ymin=ymin, ymax=ymax), 
            fill= rectangles$Cols, alpha=0.05) +
  geom_text_repel(aes(Startdate, Duration, 
                      label = gsub(".* ", "", Name),
                      alpha = Power, colour = Party),
                  min.segment.length = unit(0.1, "lines")) +
  scale_color_manual( values = brewer.pal(10, "Set3")[c(5,4)], guide = "none") +
  scale_alpha_manual(values = c("government" = 1, "shadow" = 0.5)) +
  geom_point(data = Ministers, aes(Startdate, Duration, colour = Party), alpha = 0.5) +
  scale_x_date(date_breaks = "5 years", date_labels = "%Y") +
  theme_fivethirtyeight() +
  ggtitle("Longest Survivors in a Great Office",
          subtitle = "Days in Post vs. Start Date since 1964")

extras_github <- "master/Great%20Offices/Extras.csv"
Extras <- read.csv(paste0(github, extras_github), stringsAsFactors = FALSE)
cols <- c("Name", "Office", "Duration", "Power", "Party")
Officers <- Ministers[cols]
  Officers <- rbind(Officers, Extras)
  duplicates <- which(duplicated(paste(Officers$Name, Officers$Office, Officers$Power)))
  Officers <- Officers[-duplicates,]
Offices_Number <- data.frame(table(Officers$Name))
  Offices_Number <- Offices_Number[order(-Offices_Number$Freq),]
  head(Offices_Number, n = 10)

GovOffices_Number <- data.frame(table(paste0(Officers$Name, "_", Officers$Power)))
  GovOffices_Number <- GovOffices_Number[order(-GovOffices_Number$Freq),]
  GovOffices_Number <- GovOffices_Number[-grep("shadow", GovOffices_Number$Var1),]
  GovOffices_Number$Var1 <- gsub("_.*", "", GovOffices_Number$Var1)
  head(GovOffices_Number)

  
Officers <- Officers[which(Officers$Name %in% GovOffices_Number$Var1),]
  levels(Officers$Power) <- c("shadow", "government")
ggplot(Officers, aes(x=reorder(Name, -table(Name)[Name]))) +
    geom_bar(aes(fill = Party, alpha = factor(Power, levels = c("shadow", "government")))) +
    scale_alpha_manual(values = c("government" = 1, "shadow" = 0.6), guide = guide_legend(title = "Power")) +
    scale_fill_manual( values = brewer.pal(10, "Set3")[c(5,4)], guide = "none") +
    theme_fivethirtyeight() +
    theme(axis.text.x = element_text(angle = 45, hjust = 1, size = 7)) +
    ggtitle("Holders of Most Great Offices of State",
            subtitle = "People who Have Held at Least One Governmental Office Since 1964")

Officers <- Ministers[cols]
    Officers <- rbind(Officers, Extras)
shad_rows <- which(Officers$Power == "shadow")
gov_rows <- which(Officers$Power == "government")
power_func <- function(name){
  rows <- which(Officers$Name == name)
  gov <- which(gov_rows %in% rows)
    gov_days <- sum(Officers$Duration[gov_rows[gov]])
  shad <- which(shad_rows %in% rows)
    shad_days <- sum(Officers$Duration[shad_rows[shad]])
  party <- unique(Officers$Party[rows])
  return(c(Name = name,
           Gov = gov_days,
           Shad = shad_days,
           Party = party))
}
Days <- do.call(rbind.data.frame, lapply(unique(Officers$Name), power_func))
  names(Days) <- c("Name", "Government", "Shadow", "Party")
  Days$Shadow <- as.numeric(as.character(Days$Shadow))
  Days$Government <- as.numeric(as.character(Days$Government))
selected_names <- which(Days$Name %in% c("Margaret Thatcher", "Neil Kinnock", "Roy Hattersley", "Douglas Hurd", "Tony Blair", "Gordon Brown", "James Callaghan", "Harold Wilson", "Theresa May", "David Cameron", "George Osborne", "William Hague", "Edward Heath", "John Major"))
Days_Named <- Days[selected_names,]
ggplot(Days_Named, aes(x = Government, y = Shadow, colour = Party)) +
  geom_point(data = Days) +
  geom_text_repel(label = Days_Named$Name, min.segment.length = unit(0.1, "lines")) +
  theme_fivethirtyeight() +
  scale_colour_manual( values = brewer.pal(10, "Set3")[c(5,4)], guide = "none") +
  scale_x_continuous(limits = c(0, 6000), breaks = seq(0, 6000, by = 1000)) +
  scale_y_continuous(limits = c(0, 6000), breaks = seq(0, 6000, by = 1000)) +
  ggtitle("Length of Governmental vs. Opposition Service",
          subtitle = "Total Days Holding A Shadow Great Office vs. Governmental Great Office")

library(rgdal, quietly = TRUE)
map_github <- "master/Great%20Offices/compressedWestminster.zip"
url <- paste0(github, map_github)
tmp <- tempfile( fileext = ".zip" )
download.file(url, tmp, quiet = TRUE)
unzip(tmp, exdir = tempdir()) 
map <- readOGR(dsn = tempdir(), layer = "SimplifiedWestminster", verbose = FALSE)
plot(map)