About me

• PhD in statistics from Iowa State (defended 4 months ago!)
• Maintainer of plotly's R package (for nearly 2 years!)
  • Before that: animint, LDAvis, pitchRx, rdom

About the workshop

• Focus on things that are hard to learn from documentation alone.
• Today is mainly about core ideas (& lots of mapping examples).
• Tomorrow is a bit more advanced: animation, linked views, & shiny.
• I'm hoping to save time today/tomorrow for a Q&A session.
  • Feel free to stop me at any time.
  • Feel free to ask me about personal projects during down time.

About the attendees

https://plot.ly/ggplot2/geom_density/

About the attendees, another look

https://plot.ly/r/parallel-coordinates-plot/

R wisdom

Everything that exists is an object.

Everything that happens is a function call.

-- John Chambers

R wisdom

Everything that exists is an object.

Everything that happens is a function call.

-- John Chambers

Universal truths

Use the str() function to inspect any object (View() in RStudio is a nice interactive alternative).

str(mtcars)
#> 'data.frame':    32 obs. of  11 variables:
#>  $ mpg : num  21 21 22.8 21.4 18.7 18.1 14.3 24.4 22.8 19.2 ...
#>  $ cyl : num  6 6 4 6 8 6 8 4 4 6 ...
#>  $ disp: num  160 160 108 258 360 ...
#>  $ hp  : num  110 110 93 110 175 105 245 62 95 123 ...
#>  $ drat: num  3.9 3.9 3.85 3.08 3.15 2.76 3.21 3.69 3.92 3.92 ...
#>  $ wt  : num  2.62 2.88 2.32 3.21 3.44 ...
#>  $ qsec: num  16.5 17 18.6 19.4 17 ...
#>  $ vs  : num  0 0 1 1 0 1 0 1 1 1 ...
#>  $ am  : num  1 1 1 0 0 0 0 0 0 0 ...
#>  $ gear: num  4 4 4 3 3 3 3 4 4 4 ...
#>  $ carb: num  4 4 1 1 2 1 4 2 2 4 ...

Use <- to assign value(s) to a name

nms <- names(mtcars)
nms
#>  [1] "mpg"  "cyl"  "disp" "hp"   "drat" "wt"   "qsec" "vs"   "am"   "gear"
#> [11] "carb"

The pipe operator (%>%)

Takes object on LHS and inserts into function on RHS.

library(magrittr)
mtcars %>% names()
#>  [1] "mpg"  "cyl"  "disp" "hp"   "drat" "wt"   "qsec" "vs"   "am"   "gear"
#> [11] "carb"

Makes function composition more readable

# read left-to-right
mtcars %>% names() %>% length()
#> [1] 11
# not inside out
length(names(mtcars))
#> [1] 11

R's basic data structures

A data frame holds (homo or hetero!) 1d vectors.

More than a table

What data goes into drawing this map?

library(albersusa)
usa <- usa_sf("laea")
library(dplyr)
usa %>%
 select(name, pop_2010, geometry) %>%
 View()

usa_sf("laea") %>% select(pop_2010, geometry) %>% plot()

An aside on dplyr

The R package dplyr makes common SQL-like operations fast and easy¹

Important single table operations:

• select()
• mutate()
• filter()
• arrange()
• distinct()
• summarise()

I will use dplyr sporadically through the workshop...please stop me if anything needs more explaining

library(plotly)
usa_sf <- mutate(
  usa_sf("laea"), txt = paste("The state of", name, "had \n", pop_2010, "people in 2010")
)
p <- ggplot(usa_sf) + 
  geom_sf(aes(fill = pop_2010, text = txt)) 
ggplotly(p, tooltip = "text")

Add 2 lines, & voila!

See demo("highlight-epl", package = "plotly")

ggplotly returns a plotly htmlwidget

class(p)
#> [1] "gg"     "ggplot"
gg <- ggplotly(p, tooltip = "text")
class(gg)
#> [1] "plotly"     "htmlwidget"

The htmlwidgets framework guarantees things just work in any context.¹

The htmlwidgets gallery has 85 registered widgets to date! http://gallery.htmlwidgets.org/

What happens when you print a plotly htmlwidget?

Mapping R list to JSON

barchart <- list(
  data = list(list(
    x = c("a", "b", "c"),
    y = c(1, 2, 3),
    type = "bar"
  ))
)
plotly:::to_JSON(barchart, pretty = TRUE)
#> {
#>   "data": [
#>     {
#>       "x": ["a", "b", "c"],
#>       "y": [1, 2, 3],
#>       "type": "bar"
#>     }
#>   ]
#> }

Indexing/subsetting in R

Grab a list element with $ or [[

str(barchart$data)
#> List of 1
#>  $ :List of 3
#>   ..$ x   : chr [1:3] "a" "b" "c"
#>   ..$ y   : num [1:3] 1 2 3
#>   ..$ type: chr "bar"
identical(barchart$data, barchart[["data"]])
#> [1] TRUE

There is also [, which always returns the "container"!

str(barchart["data"])
#> List of 1
#>  $ data:List of 1
#>   ..$ :List of 3
#>   .. ..$ x   : chr [1:3] "a" "b" "c"
#>   .. ..$ y   : num [1:3] 1 2 3
#>   .. ..$ type: chr "bar"

str(mtcars["vs"])
#> 'data.frame':    32 obs. of  1 variable:
#>  $ vs: num  0 0 1 1 0 1 0 1 1 1 ...
str(mtcars[["vs"]])
#>  num [1:32] 0 0 1 1 0 1 0 1 1 1 ...

Mapping R list to plotly

library(plotly)
as_widget(barchart)

PSA: use plot_ly() over as_widget()

# plot_ly() adds some useful abstractions that we'll get to later
plot_ly() %>%
  add_bars(
    x =  c("a", "b", "c"),
    y = c(1, 2, 3), 
    unsupported = "nonsense"
  )
#> Warning: 'bar' objects don't have these attributes: 'unsupported'
#> Valid attributes include:
#> 'type', 'visible', 'showlegend', 'legendgroup', 'opacity', 'name', 'uid', 'hoverinfo', 'stream', 'x', 'x0', 'dx', 'y', 'y0', 'dy', 'text', 'hovertext', 'textposition', 'textfont', 'insidetextfont', 'outsidetextfont', 'orientation', 'base', 'offset', 'width', 'marker', 'r', 't', 'error_y', 'error_x', '_deprecated', 'xaxis', 'yaxis', 'xcalendar', 'ycalendar', 'xsrc', 'ysrc', 'textsrc', 'hovertextsrc', 'textpositionsrc', 'basesrc', 'offsetsrc', 'widthsrc', 'rsrc', 'tsrc', 'key', 'set', 'frame', 'transforms', '_isNestedKey', '_isSimpleKey', '_isGraticule'

Inspect the JSON behind any widget

# In recent versions of RStudio -- gg %>% plotly_build() %>% View()
plotly_json(gg)

The data, layout, and config attributes are official plotly.js attributes covered in the figure reference.

The other attributes are unique to the R package (don't worry about them).

Modify any widget

style() modifies data attributes. layout() modifies the layout.

gg2 <- gg %>%
  style(mode = "markers+lines", traces = 2) %>%
  layout(title = "A map of 2010 population", margin = list(t = 30))

Note the modification!

# The 'x' element stores the list converted to JSON
# plotly_json() just provides a more pleasant interface to gg$x
str(gg2$x$data[[2]])

#> List of 14
#>  $ x         : num [1:238] -324546 -325004 -325571 -325589 -326462 ...
#>  $ y         : num [1:238] -110164 -119390 -130838 -131207 -149343 ...
#>  $ text      : chr "The state of Wyoming had <br /> 564358 people in 2010"
#>  $ type      : chr "scatter"
#>  $ mode      : chr "markers+lines"
#>  $ line      :List of 3
#>   ..$ width: num 1.89
#>   ..$ color: chr "rgba(89,89,89,1)"
#>   ..$ dash : chr "solid"
#>  $ fill      : chr "toself"
#>  $ fillcolor : chr "rgba(19,43,67,1)"
#>  $ hoveron   : chr "fills"
#>  $ showlegend: logi FALSE
#>  $ xaxis     : chr "x"
#>  $ yaxis     : chr "y"
#>  $ hoverinfo : chr "text"
#>  $ frame     : chr NA

Can also add data to any widget

There are a number of add_*() functions (e.g., add_bars(), add_polygons(), add_trace()).

d <- gg$x$data[[52]]
add_polygons(gg, x = d$x, y = d$y, color = I("red"), inherit = FALSE)

Resources for studying the figure reference

https://plot.ly/r/reference/

https://github.com/rreusser/plotly-doc-viewer

# In recent versions of RStudio -- View(plotly:::Schema)
schema()

Raster objects

• Raster objects are basically a matrix of color codes. These objects can be used to represent bitmap images.

m <- matrix(hcl(0, 80, seq(50, 80, 10)), nrow = 4, ncol = 5)
(r <- as.raster(m))
#>      [,1]      [,2]      [,3]      [,4]      [,5]     
#> [1,] "#C54E6D" "#C54E6D" "#C54E6D" "#C54E6D" "#C54E6D"
#> [2,] "#E16A86" "#E16A86" "#E16A86" "#E16A86" "#E16A86"
#> [3,] "#FE86A1" "#FE86A1" "#FE86A1" "#FE86A1" "#FE86A1"
#> [4,] "#FFA2BC" "#FFA2BC" "#FFA2BC" "#FFA2BC" "#FFA2BC"
plot(r)

Embedding raster objects in plotly

  plot_ly() %>%
  layout(images = list(
    source = raster2uri(r), # converts a raster object to a data URI.
    xref = "x", yref = "y", x = 0, y = 0, sizex = 1, sizey = 1,
    sizing = "stretch", xanchor = "left", yanchor = "bottom"
  ))

ggmap objects are bitmap images!

library(ggmap)
basemap <- get_map(maptype = "satellite", zoom = 8)
p <- ggmap(basemap) +
  geom_polygon(aes(x = lon, y = lat, group = plotOrder),
    data = zips, colour = "black", fill = NA) +
  ggthemes::theme_map()

Add zoom/pan/tooltips via ggplotly()

ggplotly(p)

A crossroads in ggplot2 mapping

• Going forward, geom_sf() will be the preferred way to map in ggplot2, but it is still under development.
• It will support any map projection, thanks to the magic of sf.

library(maps)
library(sf)
world1 <- st_as_sf(map('world', plot = FALSE, fill = TRUE))
ggplot() + geom_sf(data = world1)

Finding map projections

• http://www.spatialreference.org/ref/ is invaluable for finding map projections.

Mollweide projection of Canada

# http://spatialreference.org/ref/sr-org/7/proj4/
canada <- subset(world1, ID == "Canada")
canada2 <- st_transform(canada,"+proj=moll +lon_0=0 +x_0=0 +y_0=0 +ellps=WGS84 +units=m +no_defs")
ggplot() + geom_sf(data = canada2)

Now with ggplotly()

ggplotly()