##############################################
#
#   Statistics Canada data on 33 Canadian cities
#    1. minorityPercent ... visible minority %ages
#

setwd("~/Documents/Research/talks/SSC\ 2016/talk/Rcode")

library(loon)
source("minorityData.R")

# START
# Look at marginal distributions
#
# First, just for the proportion of the city population that is a "visible minority"
l_hist(minorityPercent[,"% Minority"],
       title = paste("% Minority in 33 Canadian cities"),
       xlabel="percent of population",
       linkingGroup="minority",
       yshows="density",
       showScales=TRUE
)
# END


# START
# Now show the same for the three minority groups
# having largest percentages (in decreasing order):
groups <- names(minorityPercent)[-c(12, 14, 15)]
largest3 <- names(sort(apply(minorityPercent[,groups ],
                             2, max),
                       decreasing = TRUE)
                  )[1:3]
#
for (x in largest3) {
  l_hist(minorityPercent[,x],
         title = x,
         xlabel="percent of population",
         linkingGroup="minority",
         showScales=TRUE
         )
}

#  END



# START
# Now look at the population sizes
#
pop_p <- l_plot(x=minorityPercent[,"Log_10 Total pop"],
       y=minorityPercent[,"Log_10 Minority pop"],
       title="Population sizes",
       linkingGroup="minority",
       xlabel="Log_10 total pop",
       ylabel="Log_10 minority pop",
       itemlabel=rownames(minorityPercent),
       showItemlabels=TRUE
       )
#  END


# START
# The map
# First plot the points

citiesLat <- minority[,"lat"]
citiesLong <- minority[,"long"]
p_map <- l_plot(citiesLong,citiesLat,
             xlabel="longitude", ylabel="latitude",
             linkingGroup="minority",
             itemlabel=rownames(minorityPercent),
             showItemlabels=TRUE)

#
# Get the map, add it to the plot (as a layer)
#
library(maps)
id <- l_layer(p_map,
              map("world", "Canada",
                  plot=FALSE, fill=TRUE),
              index="end")
# Rescale the plot to the size of the map
l_scaleto_layer(p_map, id)


#
# Add the text of the city names to the map

l_glyph_add_text(p_map,
                 text=row.names(minorityPercent),
                 label="city names")
# END




# START
##
#  Serial axes
#

# Back to the raw data and just rename variables
library(dplyr)
minority_star <- minority %>%
  select("SE.Asian"=Southeast.Asian,
         Chinese = Chinese,
         Japanese = Japanese,
         "S.Asian" = South.Asian,
         Other = Visible.minority.not.included.elsewhere,
         Black = Black,
         "Multiple" = Multiple.visible.minority,
         Filipino = Filipino,
         Arab = Arab,
         Korean = Korean,
         "Latino" = Latin.American)

# Get axes sequence
library(PairViz)
set.seed(1564) # otherwise axes sequence differs
CorrelationMatrix <- cor(minority_star)
VarDistance <- dist(t(minority_star))
# use minus the distance to produce the greatest
# distinction between axes
o <- order_tsp(-VarDistance)

# west to east color mapping
library(scales)
west_east_cols <- (col_numeric(c("firebrick", "orange", "sienna",
                                 "darkgreen", "steelblue", "purple"),
                               domain = NULL))(minority$long)

s <- l_serialaxes(data=minority_star, linkingGroup = "minority",
                  sequence=names(minority_star)[o],
                  showGuides=FALSE,
                  linewidth=5,
                  color=west_east_cols, showArea=FALSE,
                  itemlabel=rownames(minority), showItemlabels=TRUE)
# END





# START
## Produce grid of star glyphs
p_stars <- with(minority,
                l_plot(rank(long), rank(long),
                       title = "", color = west_east_cols,
                       showScales = FALSE, showLabels=FALSE, size=3,
                       linkingGroup = "minority",
                       itemlabel=rownames(minority), showItemlabels=TRUE))

gstars <-l_glyph_add_serialaxes(p_stars , data=minority_star,
                                sequence = names(minority_star)[o],
                                scaling = 'observation',
                                showArea=TRUE)
p_stars['glyph'] <- gstars

l_move_grid(p_stars, which='all')
## swapAxes neccessary to arrange west-east in the grid (columns)
p_stars['swapAxes'] <- TRUE
l_zoom(p_stars, .9)
#
# END
#



# START
#
# Add some glyphs to the map,
#     scale values within observations
so <- l_glyph_add_serialaxes(p_map, data=minority_star,
                             sequence = names(minority_star)[o],
                             scaling = 'observation',
                             showArea=TRUE,
                             label="stars")
p_map['glyph'] <- so

#

## Create a Worldview
p_map_ww <- l_worldview(activewidget=p_map)
#
# END

####  BACK TO SLIDES