KDnuggets Home » News » 2018 » Mar » Tutorials, Overviews » Choropleth Maps in R ( 18:n11 )

Choropleth Maps in R

Choropleth maps provides a very simple and easy way to understand visualizations of a measurement across different geographical areas, be it states or countries.

By Perceptive Analytics

If you were to compare growth rate of Indian states and present it to a bunch of people who have 15-20 seconds to look at it and infer insights from the data, what would be the right way? The best way? Would presenting the data in the traditional tabular format make sense? Or bar graphs would look better?

Bar graphs, indeed, will look better and present the data in visually appealing manner and provide a good comparison; but, will it make an impact in 15 seconds? I personally won’t be able to bring the desired outcome, moreover data for 36 states and union territories in 36 bars will make it cumbersome to scroll up and down. We have a much better alternative to table and bar charts, choropleth maps.

Choropleth maps are thematic maps in which different areas are colored or shaded in accordance with the value of a statistical variable being represented in the map. Taking an example, let’s say we were to compare population density in different states of the United States of America in a colorful manner, choropleth maps would be our best bet for representation. To sum it up, choropleth maps provides a very simple and easy way to understand visualizations of a measurement across different geographical areas, be it states or countries.

Let’s take some examples of choropleth maps and where they come handy in presenting data.

  1. Choropleth maps are widely used to represent macroeconomic variables such as GDP growth rate, population density, per-capita income, etc. on a world map and provide a proportional comparison among countries. This can also be done for states within a country.
  2. These maps can also be used to present nominal data such as gain/loss/no change in number of seats by an election party in a country.

One of the limitations of using choropleth maps is that they don’t provide details of total or absolute values. They are among the best for proportional comparison but when it comes to presenting absolute values, choropleth maps are not the right fit.

Now, let us try to see the practical implementation of choropleth maps in R. In the following code, we will try to achieve the following objectives as part of the overall implementation of the maps.

  1. Download and import the maps shape in R
  2. Creating our own dataset and representing it in the map of India
  3. Merging dataset and preparing it for visual representation
  4. Improving visualization
  5. Display external data on choropleth maps
  6. Presenting multiple maps at once

Download and import the maps share in R

There are multiple sites from where you can download shape files for free. I used this site (http://www.diva-gis.org/gdata) for downloading administrative map of India for further processing. Once you download the file, unzip the file and set your R working directory to the unzipped folder.

We will install all the necessary libraries at once and discuss one by one as we proceed along.

# Install all necessary packages and load the libraries into R

Set the working directory to the unzipped folder and use the following code to import the shape into R.

# Set working directory
states_shape = readShapeSpatial("IND_adm1.shp")
plot(states_shape, main = "Administrative Map of India")

> class(states_shape)
[1] "SpatialPolygonsDataFrame"
[1] "sp"
> names(states_shape)
[1] "ID_0"      "ISO"       "NAME_0"    "ID_1"  	"NAME_1"	"TYPE_1"	"ENGTYPE_1" "NL_NAME_1" "VARNAME_1"
> print(states_shape$ID_1)
 [1]  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
> print(states_shape$NAME_1)
[1] Andaman and Nicobar	Andhra Pradesh     	        Arunachal Pradesh       Assam              	Bihar        	     
 [6] Chandigarh         	Chhattisgarh       	Dadra and Nagar Haveli Daman and Diu      	Delhi             	
[11] Goa                	Gujarat            	Haryana            	Himachal Pradesh   	Jammu and Kashmir 	
[16] Jharkhand          	Karnataka          	Kerala             	Lakshadweep        	Madhya Pradesh    	
[21] Maharashtra        	Manipur            	Meghalaya          	Mizoram            	Nagaland          	
[26] Orissa           	        Puducherry         	Punjab             	Rajasthan          	Sikkim            	
[31] Tamil Nadu         	Telangana          	Tripura            	Uttar Pradesh      	Uttaranchal       	
[36] West Bengal       	
36 Levels: Andaman and Nicobar Andhra Pradesh Arunachal Pradesh Assam Bihar Chandigarh Chhattisgarh ... West Bengal
> plot(states_shape, main = "Administrative Map of India")


ID_1 provides a unique id for each of 36 states and union territories; while the NAME_1 provides the name of each of the states and union territories. We will be mainly using these two fields, other fields provide name of the country, code of the country and other information which separates data of one country from the other.

Alternatively, there is another function from different package which we can use to import shape into R.

States_shape2 = readOGR(".","IND_adm1")

> States_shape2<-readOGR(".","IND_adm1")
OGR data source with driver: ESRI Shapefile
Source: ".", layer: "IND_adm1"
with 36 features
It has 9 fields
Integer64 fields read as strings:  ID_0 ID_1
> class(States_shape2)
[1] "SpatialPolygonsDataFrame"
[1] "sp"
> names(States_shape2)
[1] "ID_0"  	"ISO"   	"NAME_0"	"ID_1"  	"NAME_1"	"TYPE_1"	"ENGTYPE_1" "NL_NAME_1" "VARNAME_1"
> plot(States_shape2)

In the above code “readOGR(“.”,”IND_adm1”), “.” means that the shapefile which we want to read is in our working directory; else, we would have to mention the entire path. Also, we need to mention the shapefile name without extension otherwise it will throw an error.

Creating our own dataset and representing it in the map of India

To begin with, we will create our own data for each of the 36 IDs and call it score D, a parameter which represents dancing talent of each of the states. (Please note that this score is randomly generated and does not reflect the true dancing talent :P).

# Creating our own dataset
State_count = length(states_shape$NAME_1)
score_1 = sample(100:1000, State_count, replace = T)
score_2 = runif(State_count, 1,1000)
score = score_1 + score_2
State_data = data.frame(id=states_shape$ID_1, NAME_1=states_shape$NAME_1, score)

> State_data
   id   NAME_1 	                        score
1   1	Andaman and Nicobar             558.2268
2   2   Andhra Pradesh                  961.7615
3   3  	Arunachal Pradesh              1586.5746
4   4   Assam                           281.1586
5   5   Bihar                           853.3299
6   6   Chandigarh                     1400.2554
7   7   Chhattisgarh                   1608.8069
8   8   Dadra and Nagar Haveli         1260.4761
9   9   Daman and Diu                  1195.7210
10 10   Delhi                           744.7406
11 11   Goa                            1443.5782
12 12   Gujarat                        1778.3428
13 13   Haryana                         560.5062
14 14   Himachal Pradesh                766.7788
15 15  	Jammu and Kashmir              1118.1993
16 16   Jharkhand                       901.4804
17 17   Karnataka                       520.4586
18 18   Kerala                          697.6118
19 19   Lakshadweep                    1014.7297
20 20   Madhya Pradesh                  975.1373
21 21   Maharashtra                     706.3637
22 22   Manipur                         970.6760
23 23   Meghalaya                      1182.9777
24 24   Mizoram                         986.1971
25 25   Nagaland                        942.2375
26 26   Orissa                          901.4541
27 27   Puducherry                     1754.6125
28 28   Punjab                         1570.7218
29 29   Rajasthan                      1039.7029
30 30   Sikkim                          708.4160
31 31   Tamil Nadu                      995.2757
32 32   Telangana                      1381.9686
33 33   Tripura                         659.8475
34 34   Uttar Pradesh                  1653.6564
35 35   Uttaranchal                    1138.8248
36 36   West Bengal                    1229.3981

Merging dataset and preparing it for visual representation

We will use the function fortify() of ggplot2 package to get the shape file into a data frame and then merge the data frame file and dataset together.

# Fortify file
fortify_shape = fortify(states_shape, region = "ID_1")

> fortify_shape = fortify(states_shape, region = "ID_1")
> class(fortify_shape)
[1] "data.frame"

#merge with coefficients and reorder
Merged_data = merge(fortify_shape, State_data, by="id", all.x=TRUE)
Map_plot = Merged_data[order(Merged_data$order), ]

Now, let’s create a basic visualization and see how our maps looks like.

ggplot() +
  geom_polygon(data = Map_plot,
           	aes(x = long, y = lat, group = group, fill = score),
      	     color = "black", size = 0.5) +