# Load required packages from local library into the R session.

library(tidyverse)    # ggplot2, tibble, tidyr, readr, purr, dplyr
library(lubridate)    # Dates and time.
library(magrittr)     # Pipe operator %>% %<>% %T>% equals().
library(rattle)       # comcat().
library(WDI)          # World Bank Data
library(countrycode)
library(gridExtra)

# Name of the dataset.

dsname <- "stroke"

# Identify the source location of the dataset.

dsloc <- "data"

# Construct the path to the dataset and display some if it.

dsname %>%
  paste0(".csv") %>%
  file.path(dsloc, .) %T>%
  cat("Dataset:", ., "\n\n") %T>%
  {
     paste("head", .) %>%
     system(intern=TRUE) %>%
     sub("\r", "\n", .) %>%
     print()
  } ->
dspath

dspath %>%
  read_csv2(na=".") %T>%
  glimpse() %>%
  assign(dsname, ., .GlobalEnv)

dspath %>%
  read_delim(delim=";", na=".") %T>%
  glimpse() %>%
  assign(dsname, ., .GlobalEnv)

# Prepare the dataset for usage with our template.

ds <- get(dsname)

# Review the variables to optionally normalise their names.

names(ds)

# Normalise the variable names.

names(ds) %<>% normVarNames() %T>% print()

# Confirm the results are as expected.

glimpse(ds)


ds$died %<>% dmy() %T>% {class(.) %>% print()}
ds$dstr %<>% dmy() %T>% {class(.) %>% print()}

glimpse(ds)

(ct <- Sys.time())
class(ct)
str(ct)
unclass(ct)

(ct <- as.POSIXlt(ct))
class(ct)
str(ct)
unclass(ct)

format(Sys.time(), "%a %d %b %Y %H:%M:%S %Z")

ds$lived <- ds$died - ds$dstr
head(ds$lived)
class(ds$lived)

ds$lived <- ds$died - ds$dstr
head(ds$lived)
class(ds$lived)

as.integer(ds$lived[1])/60/60/24

units(ds$lived)
units(ds$lived) <- "days"
units(ds$lived)
head(ds$lived)

ds$interval <- with(ds, interval(dstr, died))
head(ds$interval)
class(ds$interval)
max(ds$interval, na.rm=TRUE)
min(ds$interval, na.rm=TRUE)

head(as.duration(ds$interval))

ds %>%
  mutate(weekday=wday(dstr, label=TRUE, abbr=FALSE)) %>%
  ggplot(aes(weekday)) +
  geom_bar(colour="white", fill="lightblue") +
  labs(title="Day of Week of Incidence of Stroke", x="Weekday", y="Count")

ds %>%
  mutate(mday=mday(dstr)) %>%
  ggplot(aes(mday)) +
  geom_bar(colour="white", fill="orange") +
  labs(title="Day of Month of Incidence of Stroke", x="Day of Month")

data.frame(L1=100+c(0, cumsum(runif(99, -20, 20))),
           L2=150+c(0, cumsum(runif(99, -10, 10))),
           Date=seq.Date(as.Date("2000-01-01"), 
                         by="1 month", length.out=100)) %>%
  melt(id="Date") %>%
  ggplot(aes(x=Date, y=value, colour=variable)) +
  geom_line() +
  ylab("Observation") +
  labs(colour="Location")

(meta.data <- WDIsearch("Fertility rate", field="name", short=FALSE))
(indicators <- meta.data[1:2, 1])
countries <- c("United States", "Britain", "India", "China", "Australia")
(iso2char <- countrycode(countries, "country.name", "iso2c"))
(wdids <- WDI(iso2char, meta.data[1:2,1], start=2001, end=2011))

plots <- lapply(indicators, function(nm)
{ 
  p <- ggplot(wdids, aes(x=year, y=wdids[,nm], group=country, color=country),
              environment=environment())
  p <- p + geom_line(size=1)
  p <- p + scale_x_continuous(name="Year", breaks=c(unique(wdids[,"year"])))
  p <- p + scale_y_continuous(name=nm)
  p <- p + scale_linetype_discrete(name="Country")
  p <- p + theme(legend.title=element_blank())
  p <- p + labs(title=paste(meta.data[meta.data[,1]==nm, "name"], "\n"))
})

do.call(grid.arrange, plots)

vars <- c("Date", "MinTemp", "MaxTemp", "Sunshine", "Rainfall", "Evaporation")
ds <- weather[vars]

ds$Sunshine <- ds$Sunshine * 60

ds$CumRainfall <- cumsum(ds$Rainfall)
ds$CumEvaporation <- cumsum(ds$Evaporation)

dsm <- melt(ds, id="Date")

g <- ggplot(dsm, aes(x=Date, y=value, colour=variable))
g <- g + geom_point()
print(g)            

g <- ggplot(dsm, aes(x=Date, y=value, colour=variable))
g <- g + geom_point()
g <- g + scale_y_log10()
print(g)            

asinh_trans <- function() trans_new(name="asinh",
                                    transform=asinh,
                                    inverse=sinh)
g <- ggplot(dsm, aes(x=Date, y=value, colour=variable))
g <- g + geom_point()
g <- g + scale_y_continuous(trans="asinh")
print(g)  

g <- ggplot(dsm, aes(x=Date, y=value, colour=variable))
g <- g + geom_point()
g <- g + scale_y_continuous(trans="asinh",
                            limits=c(-1e4, 1e4))
print(g)  

g <- ggplot(dsm, aes(x=Date, y=value, colour=variable))
g <- g + geom_point()
g <- g + scale_y_continuous(trans="asinh",
                            breaks=c(-10, 0, 10, 1e2, 1e3))
print(g)  

g <- ggplot(dsm, aes(x=Date, y=value, colour=variable))
g <- g + geom_point()
g <- g + scale_y_continuous(trans="asinh",
                            breaks=c(-10, 0, 10, 1e2, 1e3),
                            labels=c("-10", "0", "10", "100", "1K"))
print(g)  

g <- ggplot(dsm, aes(x=Date, y=value, colour=variable))
g <- g + geom_line()
g <- g + scale_y_continuous(trans="asinh",
                            breaks=c(-10, 0, 10, 1e2, 1e3),
                            labels=c("-10", "0", "10", "100", "1K"))
print(g)  

draw.lines <- c("CumRainfall", "CumEvaporation")

g <- ggplot(dsm, aes(x=Date, y=value, colour=variable))
g <- g + geom_point(data=subset(dsm, !variable %in% draw.lines))
g <- g + geom_line(data=subset(dsm, variable %in% draw.lines))
g <- g + scale_y_continuous(trans="asinh",
                            breaks=c(-10, 0, 10, 1e2, 1e3),
                            labels=c("-10", "0", "10", "100", "1K"))
print(g)  

events <- as.Date(c("2007-12-25", "2008-03-22"))

g <- ggplot(dsm, aes(x=Date, y=value, colour=variable))
g <- g + geom_point(data=subset(dsm, !variable %in% draw.lines))
g <- g + geom_line(data=subset(dsm, variable %in% draw.lines))
g <- g + scale_y_continuous(trans="asinh",
                            breaks=c(-10, 0, 10, 1e2, 1e3),
                            labels=c("-10", "0", "10", "100", "1K"))
g <- g + geom_vline(xintercept=as.numeric(events), linetype=3)
g <- g + annotate("text", events[1], -9, label="Christmas", size=3, colour="blue")
g <- g + annotate("text", events[2], -9, label="Easter", size=3, colour="purple")
print(g)  

g <- ggplot(dsm, aes(x=Date, y=value, colour=variable))
g <- g + geom_point(data=subset(dsm, !variable %in% draw.lines))
g <- g + geom_line(data=subset(dsm, variable %in% draw.lines))
g <- g + scale_y_continuous(trans="asinh",
                            breaks=c(-10, 0, 10, 1e2, 1e3),
                            labels=c("-10", "0", "10", "100", "1K"))
g <- g + geom_vline(xintercept=as.numeric(events), linetype=3)
g <- g + annotate("text", events[1], -9, label="Christmas", size=3, colour="blue")
g <- g + annotate("text", events[2], -9, label="Easter", size=3, colour="purple")
g <- g + labs(title=sprintf("Weather pattern for %s", weather$Location[1]))
g <- g + theme(legend.direction="horizontal", legend.position="bottom")
print(g)