Heatmap plot by value using ggmap

Question

I am attempting to use ggmap to look at education scores by school. I created a coordinate list of all the schools and the individual student scores like so:

     score      lat       lon
3205    45 28.04096 -82.54980
8275    60 27.32163 -80.37673
4645    38 27.45734 -82.52599
8962    98 26.54113 -81.84399
9199    98 27.88948 -82.31770
340     53 26.36528 -81.79639

I first used the pattern from most of the tutorials that I worked through: http://journal.r-project.org/archive/2013-1/kahle-wickham.pdf http://www.geo.ut.ee/aasa/LOOM02331/heatmap_in_R.html

library(ggmap)
library(RColorBrewer)

MyMap <- get_map(location = "Orlando, FL", 
                 source = "google", maptype = "roadmap", crop = FALSE, zoom = 7)

YlOrBr <- c("#FFFFD4", "#FED98E", "#FE9929", "#D95F0E", "#993404")

ggmap(MyMap) +
stat_density2d(data = s_rit, aes(x = lon, y = lat, fill = ..level.., alpha = ..level..),
               geom = "polygon", size = 0.01, bins = 16) +
scale_fill_gradient(low = "red", high = "green") +
scale_alpha(range = c(0, 0.3), guide = FALSE)

In the first plot, the graphics look great but it doesn't take the score into account.

In order to incorporate the score variable, I used this example Density2d Plot using another variable for the fill (similar to geom_tile)?:

ggmap(MyMap) %+% s_rit +
  aes(x = lon, y = lat, z = score) +
  stat_summary2d(fun = median, binwidth = c(.5, .5), alpha = 0.5) +
  scale_fill_gradientn(name = "Median", colours = YlOrBr, space = "Lab") +
  labs(x = "Longitude", y = "Latitude") +
  coord_map()

It colours by value, but it doesn't have the look of the first. The square boxes are clunky and arbitrary. Adjusting the size of the box does not help. The dispersion of the first heatmap is preferred. Is there a way to blend the look of the first graph with the value-based plot of the second?

Data

s_rit <- structure(list(score = c(45, 60, 38, 98, 98, 53, 90, 42, 96, 
45, 89, 18, 66, 2, 45, 98, 6, 83, 63, 86, 63, 81, 70, 8, 78, 
15, 7, 86, 15, 63, 55, 13, 83, 76, 78, 70, 64, 88, 61, 78, 4, 
7, 1, 70, 88, 58, 70, 58, 11, 45, 28, 42, 45, 73, 85, 86, 25, 
17, 53, 95, 49, 80, 70, 35, 94, 61, 39, 76, 28, 1, 18, 93, 73, 
67, 56, 38, 45, 66, 18, 76, 91, 76, 52, 60, 2, 38, 73, 95, 1, 
76, 6, 25, 76, 81, 35, 49, 85, 55, 66, 90), lat = c(28.040961, 
27.321633, 27.457342, 26.541129, 27.889476, 26.365284, 28.555024, 
26.541129, 26.272728, 28.279994, 27.889476, 28.279994, 26.6674, 
26.272728, 25.776045, 26.541129, 30.247658, 26.365284, 25.450123, 
27.889476, 26.541129, 27.264513, 26.718652, 28.044369, 28.251435, 
27.264513, 26.272728, 26.272728, 28.040961, 30.312239, 27.889476, 
26.541129, 26.6674, 27.321633, 26.365284, 28.279994, 26.718652, 
30.23286, 28.040961, 30.193704, 30.312239, 28.044369, 27.457342, 
25.450123, 30.23286, 30.312239, 30.193704, 28.279994, 30.247658, 
26.541129, 26.365284, 28.279994, 27.321633, 25.776045, 26.272728, 
30.23286, 30.312239, 26.718652, 26.541129, 25.450123, 28.251435, 
28.185751, 25.450123, 28.040961, 27.321633, 28.279994, 27.321633, 
27.321633, 27.321633, 28.279994, 26.718652, 28.362308, 27.264513, 
26.365284, 28.279994, 30.23286, 25.450123, 28.362308, 25.450123, 
25.776045, 30.193704, 28.251435, 27.457342, 27.321633, 28.185751, 
27.457342, 27.889476, 26.541129, 26.541129, 30.23286, 30.312239, 
26.718652, 25.450123, 26.139258, 28.040961, 30.23286, 26.718652, 
28.185751, 28.044369, 28.555024), lon = c(-82.5498, -80.376729, 
-82.525985, -81.843986, -82.317701, -81.796389, -81.276464, -81.843986, 
-80.207508, -81.331178, -82.317701, -81.331178, -80.072089, -80.207508, 
-80.199437, -81.843986, -81.808664, -81.796389, -80.433557, -82.317701, 
-81.843986, -80.432125, -80.091078, -82.394639, -81.490407, -80.432125, 
-80.207508, -80.207508, -82.5498, -81.575916, -82.317701, -81.843986, 
-80.072089, -80.376729, -81.796389, -81.331178, -80.091078, -81.585975, 
-82.5498, -81.579846, -81.575916, -82.394639, -82.525985, -80.433557, 
-81.585975, -81.575916, -81.579846, -81.331178, -81.808664, -81.843986, 
-81.796389, -81.331178, -80.376729, -80.199437, -80.207508, -81.585975, 
-81.575916, -80.091078, -81.843986, -80.433557, -81.490407, -81.289394, 
-80.433557, -82.5498, -80.376729, -81.331178, -80.376729, -80.376729, 
-80.376729, -81.331178, -80.091078, -81.428494, -80.432125, -81.796389, 
-81.331178, -81.585975, -80.433557, -81.428494, -80.433557, -80.199437, 
-81.579846, -81.490407, -82.525985, -80.376729, -81.289394, -82.525985, 
-82.317701, -81.843986, -81.843986, -81.585975, -81.575916, -80.091078, 
-80.433557, -80.238901, -82.5498, -81.585975, -80.091078, -81.289394, 
-82.394639, -81.276464)), .Names = c("score", "lat", "lon"), row.names = c(3205L, 
8275L, 4645L, 8962L, 9199L, 340L, 5381L, 8998L, 5476L, 4956L, 
9256L, 4940L, 6681L, 5586L, 1046L, 9017L, 1878L, 323L, 4175L, 
9236L, 8968L, 6885L, 5874L, 9412L, 6434L, 7168L, 5420L, 5680L, 
3202L, 1486L, 9255L, 9009L, 6833L, 8271L, 261L, 5024L, 8028L, 
1774L, 3329L, 1824L, 1464L, 9468L, 4643L, 4389L, 1506L, 1441L, 
1826L, 4968L, 1952L, 8803L, 339L, 4868L, 8266L, 1334L, 5483L, 
1727L, 1389L, 7944L, 8943L, 4416L, 6440L, 526L, 4478L, 3117L, 
8308L, 4891L, 8290L, 8299L, 8233L, 4848L, 7922L, 5795L, 6971L, 
179L, 4990L, 1776L, 4431L, 5718L, 4268L, 1157L, 1854L, 6433L, 
4637L, 8194L, 560L, 4694L, 9274L, 8903L, 8877L, 1586L, 1398L, 
5865L, 4209L, 6075L, 3307L, 1634L, 8108L, 514L, 9453L, 5210L), class = "data.frame")

Yes, I'd like the map colors to be greener where there are higher scores. Someone should be able to look at the map and tell where the scores are best and worst — Pierre L, Aug 21 '15 at 20:35
In a question with a similar data structure (lat, lon, value (rain / score)) I used interpolation with `akima` (as also mentioned by @nongkrong), then `ggmap` and `geom_tile` in [my answer](http://stackoverflow.com/questions/19339296/plotting-contours-on-an-irregular-grid/19339663#19339663). You may check if it suits your needs. — Henrik, Aug 21 '15 at 21:10

hrbrmstr · Accepted Answer · 2015-08-21T21:37:47.430

I'd like to suggest an alternate way of visualizing the distribution of scores (in general) and the median outcomes of each school. It might be better (I don't really know your data or overall problem statement) to show the distribution of scores themselves by various levels (0-10, 10-20, etc) separately then show a view of the actual median rankings per school. Something like this:

library(ggplot2)
library(ggthemes)
library(viridis) # devtools::install_github("sjmgarnier/viridis)
library(ggmap)
library(scales)
library(grid)
library(dplyr)
library(gridExtra)

dat$cut <- cut(dat$score, breaks=seq(0,100,11), labels=sprintf("Score %d-%d",seq(0, 80, 10), seq(10,90,10)))

orlando <- get_map(location="orlando, fl", source="osm", color="bw", crop=FALSE, zoom=7)

gg <- ggmap(orlando)
gg <- gg + stat_density2d(data=dat, aes(x=lon, y=lat, fill=..level.., alpha=..level..),
                          geom="polygon", size=0.01, bins=5)
gg <- gg + scale_fill_viridis()
gg <- gg + scale_alpha(range=c(0.2, 0.4), guide=FALSE)
gg <- gg + coord_map()
gg <- gg + facet_wrap(~cut, ncol=3)
gg <- gg + labs(x=NULL, y=NULL, title="Score Distribution Across All Schools\n")
gg <- gg + theme_map(base_family="Helvetica")
gg <- gg + theme(plot.title=element_text(face="bold", hjust=1))
gg <- gg + theme(panel.margin.x=unit(1, "cm"))
gg <- gg + theme(panel.margin.y=unit(1, "cm"))
gg <- gg + theme(legend.position="right")
gg <- gg + theme(strip.background=element_rect(fill="white", color="white"))
gg <- gg + theme(strip.text=element_text(face="bold", hjust=0))
gg

median_scores <- summarise(group_by(dat, lon, lat), median=median(score))
median_scores$school <- sprintf("School #%d", 1:nrow(median_scores))

gg <- ggplot(median_scores)
gg <- gg + geom_segment(aes(x=reorder(school, median), 
                            xend=reorder(school, median), 
                            y=0, yend=median), size=0.5)
gg <- gg + geom_point(aes(x=reorder(school, median), y=median))
gg <- gg + geom_text(aes(x=reorder(school, median), y=median, label=median), size=3, hjust=-0.75)
gg <- gg + scale_y_continuous(expand=c(0, 0), limits=c(0, 100))
gg <- gg + labs(x=NULL, y=NULL, title="Median Score Per School")
gg <- gg + coord_flip()
gg <- gg + theme_tufte(base_family="Helvetica")
gg <- gg + theme(axis.ticks.x=element_blank())
gg <- gg + theme(axis.text.x=element_blank())
gg <- gg + theme(plot.title=element_text(face="bold", hjust=1))
gg_med <- gg

# tweak hjust and potentially y as needed
median_scores$hjust <- 0
median_scores[median_scores$school=="School #10",]$hjust <- 1.5
median_scores[median_scores$school=="School #8",]$hjust <- 0
median_scores[median_scores$school=="School #9",]$hjust <- 1.5

gg <- ggmap(orlando)
gg <- gg + geom_text(data=median_scores, aes(x=lon, y=lat, label=gsub("School ", "", school)), 
                     hjust=median_scores$hjust, size=3, face="bold", color="darkblue")
gg <- gg + coord_map()
gg <- gg + labs(x=NULL, y=NULL, title=NULL)
gg <- gg + theme_map(base_family="Helvetica")
gg_med_map <- gg

grid.arrange(gg_med_map, gg_med, ncol=2)

Adjust the school labels on the map as needed.

That should help show whatever geographic causality (or lack of) you're trying to show.

Thanks for the response. But if we need to plot the map as @fortune_p wants ? I face exactly the same problem that is described in this question. — Kumpelka, Oct 26 '17 at 13:59

Heatmap plot by value using ggmap

1 Answers1

Linked