Visualizing Spatial Data

The Spatially Enabled Dataframe has a plot() method that uses a syntax and symbology similar to matplotlib for visualizing features on a map. With this functionality, you can easily visualize aspects of your data both on a map and on a matplotlib chart using the same symbology!

Some unique characteristics of working with the visualization capabalities on the SDF:

  • Uses Pythonic syntax
  • Uses the same syntax as visualizing charts on Pandas DataFrames
  • Uses symbology familiar to users of matplotlib
  • Works on features and attributes simultaneously, eliminating to a great extent the need to iterate over all features (rows)
  • Handles reading and writing to multiple formats aiding data conversion

Checkout the Introduction to Spatially Enabled DataFrame guide to learn how to create a Spatially Enabled DataFrame.


Let us read a census data on major cities and load that into a Spatially Enabled DataFrame

In [11]:
from arcgis import GIS
gis = GIS()
# create an anonymous connection to ArcGIS Online and get a public item
item = gis.content.get("85d0ca4ea1ca4b9abf0c51b9bd34de2e")
flayer = item.layers[0]

# Specify a SQL query and get a sub-set of the original data as a DataFrame
df = flayer.query(where="AGE_45_54 < 1500").sdf

# Visualize the top 5 records
0 1413 1381 1106 2138 1815 1411 979 1557 525 307 ... 0468080 14287 14980 6 1074 {"x": -12768343.256613126, "y": 3842463.708135... AZ 04 261 9196
1 727 738 677 1380 1185 1333 1087 740 661 444 ... 0602042 9932 10239 6 2056 {"x": -13613950.337588644, "y": 4931686.754090... CA 06 267 8273
2 593 511 2323 2767 746 127 34 1229 4 2 ... 0610561 10616 11869 6 2558 {"x": -13066582.116550362, "y": 3925650.676616... CA 06 296 7530
3 888 988 900 1729 1479 1443 959 766 514 280 ... 0613560 10866 11195 6 761 {"x": -13123874.446103057, "y": 4044249.710416... CA 06 86 5898
4 1086 1228 1013 1822 1759 1478 1112 925 687 477 ... 0614974 12823 13009 6 1763 {"x": -13151212.145276317, "y": 4027601.332347... CA 06 88 6930

5 rows × 51 columns

Note: If the above cell looks new to you, please checkout the Introduction to Spatially Enabled DataFrame guide.

Let us create a map of the United States

In [2]:
m1 = GIS().map('United States')

map of US diamonds

In [3]:
m1.zoom = 4 = [39,-98]

Plotting the DataFrame

You can quickly visualize the points by calling the plot() method off the DataFrame's spatial accessor and passing the map you created above.

In [4]:
df.spatial.plot(map_widget= m1)

You can customize the symbols, their color, shape, border etc. like you would in a typical matplotlib plot. The rest of this guide talks about such customizations and suggestions to visualize your spatial and non-spatial data.

The code below plots the same set of points on a new map using a common structure used amongst many different Python packages for defining symbology. It is built off of the matplotlib libraries for simple, straightforward plotting. We'll explain some of the parameters below, and the plot() API Reference outlines more options.

In [5]:
m2= GIS().map('United States', zoomlevel=4)

map of US diamonds

In [6]:[39,-98]
            symbol_style='d', # d - for diamonds

If this type of plotting is new to you, this is a good time to checkout matplotlib.pyplot documentation and Pandas dataframe plot documentation to get an idea.

Understanding renderers

In the snippets above, you visualized all the features with the same symbology. While this helps understand the geographic distribution of your data, your Spatially Enabled DataFrame object can do much more. For instance, you can group the features based on certain attributes that you choose and symbolize each group by a certain color or size or both. To help with this, let us understand the concept of renderers.

Renderers define how to visually represent a feature layer by defining symbols to represent individual features. The SeDF provides you with functionality to control the way features appear by choosing the symbol the renderer uses.

Previous versions of the ArcGIS API for Python provided a method to specify a renderer manually, but you had to know details about the renderer before you drew your data. The map.add_layer() method did not provide access to all options avialble for rendering datasets. The new visualization capabilities provided by the SEDF allow you to draw spatial data quickly and easily with access to more rendering options.

Supported renderers

The renderering options below are documented in further detail in the DatFrame Reference. At a high level, you have the following renderers:

Renderer Syntax Explanation
Simple 's' renders using one symbol only (the examples in quickstart above)
Unique 'u' renders each unique value with a different symbol. Suitable for categorical columns
Unique 'u-a' renders each unique value with a different symbol using arcade expressions. Suitable for categorical columns
Class breaks 'c' renders each group of values with a different color or size. Suitable for numerical columns
Heatmap 'h' renders density of point data as raster of varying colors

Visualizing unique values

Let us explore the major cities census data further and explore if there are any categorical columns

In [7]:
Index(['AGE_10_14', 'AGE_15_19', 'AGE_20_24', 'AGE_25_34', 'AGE_35_44',
       'AGE_45_54', 'AGE_55_64', 'AGE_5_9', 'AGE_65_74', 'AGE_75_84',
       'AGE_85_UP', 'AGE_UNDER5', 'AMERI_ES', 'ASIAN', 'AVE_FAM_SZ',

The ST column contains state names in abbreviation is a good candidate. We could symbolize the points such that all which fall under the same state get the same symbol:

In [8]:
df[['ST', 'NAME']].head()
0 AZ Somerton
1 CA Anderson
2 CA Camp Pendleton South
3 CA Citrus
4 CA Commerce
In [9]:
gis = GIS()
m3 ='Reno, NV', zoomlevel=4)

unique value renderer

In [10]:
df.spatial.plot(map_widget = m3,
                renderer_type='u', # specify the unique value renderer using its notation 'u'
                col='ST'  # column to get unique values from

By default, the API picks a unique color from the jet colormap for each category (US State names in this case) and assigns it. You can override this by choosing your own colormap (prism for instance) and customizing the symbol style, symbol type, outline color, thickness, transparency etc. Keep reading to learn more about these customizations.

It is a good cartographic practice to not have too many unique values in your legend. Human eyes can only preceive so many colors and there are only a limited number of colors to choose from in a given color map before the colors become indistinguishable.

As a safe practice, the Python API caps the maximum number of unique values to 256, the remaining are generalized and plotted with a default color. If you have several unique values, you should consider picking a different column that better represents your data or consider using a different renderer.

Visualizing Unique Values with Arcade Expressions

Arcade is an expression language that can be used across the ArcGIS Platform. Whether writing simple scripts to control how features are rendered, or expressions to control label text, Arcade provides a simple scripting syntax to deliver these capabilities.

In the sense of visualization, Arcade expressions are used to create rich and dynamic symbology. This example will follow the JavaScript guide.

Obtain the Data

Access the enterprise to gain access to the FeatureLayer information and query it into a Spatially Enabled DataFrame.

In [3]:
item = gis.content.get("8444e275037549c1acab02d2626daaee")
flayer = item.layers[0]
df = flayer.query().sdf
In [4]:
fset = flayer.query()
In [5]:
from arcgis.geometry import Geometry
g = Geometry(fset.features[0].geometry)

Write out the Arcade Expressions and Stops

Arcade expressions require the stops to be manually provided. In this case, we first create an opacity visual variable based on a percent of dominant parties in registered citizens.

In [6]:
opacity_expression = ("var republican = $feature.MP06025a_B;var democrat = $feature.MP06024a_B;"
                      "var independent = $feature.MP06026a_B;var parties = [republican, democrat, independent];"
                      "var total = Sum(parties);var max = Max(parties);return (max / total) * 100;")
opacity_stops = [
    { "value": 33, "transparency": 0.05 * 255, "label": "< 33%" },
    { "value": 44, "transparency": 1.0 * 255, "label": "> 44%" }]

Next we develop another Arcade expression to obtain the majority party in a given county.

In [7]:
arcade_expression = ("var republican = $feature.MP06025a_B;var democrat = $feature.MP06024a_B;"
                     "var independent = $feature.MP06026a_B;var parties = [republican, democrat, independent];"
                     "return Decode( Max(parties),republican, 'republican', democrat, 'democrat',independent, "
                     "'independent','n/a' );")
uv = [{"label":"Democrat","symbol":{"type":"esriSFS","color":[0,195,255,255],
      {"label":"Independent/other party","symbol":{"type":"esriSFS","color":[250,255,0,255],
In [8]:
gis = GIS()
m3_ua ='United States', zoomlevel=4)

Visualize the Data

Provide the color scheme, and the arcade_expression to render the data in a dynamic/rich form.

In [9]:
                renderer_type='u-a',  # 'u-a' stands for uniqe value with arcade expression

Visualizing classes with different colors

Often, you may want to classify the numerical values in your data into groups and visualize them on a map. You can accomplish this with a class break renderer which splits your data into specific number of groups and uses color to differentiate each group. You can choose the algorithm that performs the class splits or go with the defualt.

Let us visualize the same major cities point dataset using its POPULATION column.

In [12]:
df[['ST', 'NAME', 'POPULATION']].head()
0 AZ Somerton 14980
1 CA Anderson 10239
2 CA Camp Pendleton South 11869
3 CA Citrus 11195
4 CA Commerce 13009
In [14]:
m4 ='USA', zoomlevel=4)

class color renderer

In [15]:
               renderer_type='c',  # for class breaks renderer
               method='esriClassifyNaturalBreaks',  # classification algorithm
               class_count=20,  # choose the number of classes
               col='POPULATION',  # numeric column to classify
               cmap='prism',  # color map to pick colors from for each class
               alpha=0.7  # specify opacity

You can take this further by plotting a histogram of the same POPULATION column and try to match the colors. To get the class breaks for the above map, you can inspect the layer definition from MapView object as shown below:

In [25]:
dict_keys(['type', 'visualVariables', 'rotationType', 'minValue', 'field', 'defaultSymbol', 'defaultLabel', 'classificationMethod', 'classBreakInfos'])
In [26]:
class_breaks = m4.layers[0].layer.layerDefinition.drawingInfo.renderer.classBreakInfos

We can loop through each of the class breaks and obtain the min, max values and the colors for each class.

In [29]:
cbs_list = []
cmap_list = []
for cb in class_breaks:
    print(cb.description)  # print the class break labels
    cmap_list.append([x/255.0 for x in cb.symbol.color])
7809.0 - 10844.421052631578
10844.421052631578 - 13879.842105263157
13879.842105263157 - 16915.263157894737
16915.263157894737 - 19950.684210526313
19950.684210526313 - 22986.105263157893
22986.105263157893 - 26021.526315789473
26021.526315789473 - 29056.94736842105
29056.94736842105 - 32092.36842105263
32092.36842105263 - 35127.78947368421
35127.78947368421 - 38163.21052631579
38163.21052631579 - 41198.63157894737
41198.63157894737 - 44234.05263157895
44234.05263157895 - 47269.47368421053
47269.47368421053 - 50304.8947368421
50304.8947368421 - 53340.31578947368
53340.31578947368 - 56375.73684210526
56375.73684210526 - 59411.15789473684
59411.15789473684 - 62446.57894736842
62446.57894736842 - 65482.0
65482.0 - 65482.0

Now that we have the individual class splits and the colors, we can plot a histogram using the same breaks and colors.

In [30]:
import matplotlib.pyplot as plt
# build a histogram for the same class breaks
n, bins, patches = plt.hist(df['POPULATION'], bins=cbs_list)

# apply the same color for each class to match the map
idx = 0
for c, p in zip(bins, patches):
    plt.setp(p, 'facecolor', cmap_list[idx])

plt.title('Histogram of POPULATION column')
Text(0.5,1,'Histogram of POPULATION column')

The shape and color of the bars in the histogram match with the number and color of major cities drawn on the map.

Thus using the Spatially Enabled DataFrame we can easily visualize both the spatial and numeric distribution of data using the same symbology.

Colormaps and Colors

The examples above use the default colormap jet. However you can customize them by using the colors paramater. You can specify colors as:

Color Array

RGB + Alpha values can be used to create colors for symbols called in the plot method. RGB stands for red, green, and blue respectively. Each RGB value is a value between 0-255, and the alpha value is a number between 0-255.

Example to produce :

color = [255,0,100,1]

The above example produces a purplish color. Many websites provide details about using colors. For example, see here for a color codes chart.

Color Maps

A color map is a collection of string values that can be given to generate a series of related colors from a defined set.

Color maps can be viewed here:

Color Map Helpers

To better understand the syntax for each input type, the ArcGIS API for Python provides some helper functions:

In [31]:
from arcgis.mapping import display_colormaps

The display_colormaps function provides a quick, easy way to visualize the pre-defined set of colormaps you can use.

In [32]:

You can retrieve a list of colormaps as well:

In [33]:
from arcgis.mapping import symbol

colormaps = symbol.ALLOWED_CMAPS
for a,b,c,d,e in zip(colormaps[::5], colormaps[1::5], colormaps[2::5], colormaps[3::5], colormaps[4::5]):
Accent              Accent_r            Blues               Blues_r             BrBG
BrBG_r              BuGn                BuGn_r              BuPu                BuPu_r
CMRmap              CMRmap_r            Dark2               Dark2_r             GnBu
GnBu_r              Greens              Greens_r            Greys               Greys_r
OrRd                OrRd_r              Oranges             Oranges_r           PRGn
PRGn_r              Paired              Paired_r            Pastel1             Pastel1_r
Pastel2             Pastel2_r           PiYG                PiYG_r              PuBu
PuBuGn              PuBuGn_r            PuBu_r              PuOr                PuOr_r
PuRd                PuRd_r              Purples             Purples_r           RdBu
RdBu_r              RdGy                RdGy_r              RdPu                RdPu_r
RdYlBu              RdYlBu_r            RdYlGn              RdYlGn_r            Reds
Reds_r              Set1                Set1_r              Set2                Set2_r
Set3                Set3_r              Spectral            Spectral_r          Wistia
Wistia_r            YlGn                YlGnBu              YlGnBu_r            YlGn_r
YlOrBr              YlOrBr_r            YlOrRd              YlOrRd_r            afmhot
afmhot_r            autumn              autumn_r            binary              binary_r
bone                bone_r              brg                 brg_r               bwr
bwr_r               cool                cool_r              coolwarm            coolwarm_r
copper              copper_r            cubehelix           cubehelix_r         flag
flag_r              gist_earth          gist_earth_r        gist_gray           gist_gray_r
gist_heat           gist_heat_r         gist_ncar           gist_ncar_r         gist_rainbow
gist_rainbow_r      gist_stern          gist_stern_r        gist_yarg           gist_yarg_r
gnuplot             gnuplot2            gnuplot2_r          gnuplot_r           gray
gray_r              hot                 hot_r               hsv                 hsv_r
inferno             inferno_r           jet                 jet_r               magma
magma_r             nipy_spectral       nipy_spectral_r     ocean               ocean_r
pink                pink_r              plasma              plasma_r            prism
prism_r             rainbow             rainbow_r           seismic             seismic_r
spring              spring_r            summer              summer_r            terrain
terrain_r           viridis             viridis_r           winter              winter_r

You can enter a list of color ramp names as input to the display_colormaps function to filter the output:

In [34]:
display_colormaps(['Greens_r', 'PRGn', 'Dark2', 'Set1'])

The examples above used jet which is the default and some used prism both of which can be seen in the list above. You can customize your charts by using a different colormap of your choice.

Symbology for Simple Renderers

The ArcGIS API for Python provides you the ability to set symbol types so you control data appearance. The show_styles function in the arcgis.mapping module assists you with the syntax to define symbols.

Getting the different symbol styles

In [35]:
from arcgis.mapping import show_styles
In [36]:
# get the styles that are relevant to the current geometry type (points)
show_styles(df.spatial.geometry_type[0])  # the DataFrame in this example is of Point type
0 o Circle (default)
1 + Cross
2 d Diamond
3 s Square
4 x X
In [37]:
0 \ Backward Diagonal
1 / Forward Diagonal
2 | Vertical Bar
3 - Horizontal Bar
4 x Diagonal Cross
5 + Cross
6 s Solid Fill (default)
In [38]:
0 s Solid (default)
1 - Dash
2 -. Dash Dot
3 -.. Dash Dot Dot
4 . Dot
5 -- Long Dash
6 --. Long Dash Dot
7 n Null
8 s- Short Dash
9 s-. Short Dash Dot
10 s-.. Short Dash Dot Dot
11 s. Short Dot

For instance, you can represent the same set of points using square symbols instead of default circle using the snippet shown below:

In [ ]:
m7 = GIS().map('United States', 4)

map of US squares

In [ ]: = [39, -98]
        symbol_style='s',  # for square
        cmap='Greens_r',   # color map is Greens_r for green reversed
        cstep=35,   # the individual color in the colormap

Visualizing line features using simple symbols

Let us search for USA freeway layer and visualize it by looping through the different line symbols

In [ ]:
search_result ='title:USA freeway system AND owner:esri', 
                                  item_type = 'Feature Layer')
freeway_item = search_result[0]
In [ ]:
freeway_sdf = freeway_item.layers[0].query().sdf
In [ ]:
m8 ='USA', 10)

In [ ]:[34.05,-118.2]
In [ ]:
import time
for row, style in show_styles('Line').iterrows():
    time.sleep(1)  # sleep for 1 second so animation can finish
    print(style['ESRI_STYLE'] + " : " + style['MARKER'])
                     cmap='Spectral',  # user a different color map

Visualizing area features using different symbols

In [ ]:
from arcgis.features import FeatureLayer
fl = FeatureLayer("")
county_sdf = fl.query("STATE_NAME='Washington'", out_sr=4326).sdf
In [ ]:
m9 ='Seattle, WA', zoomlevel=6)

In [ ]:
for idx, style in show_styles('Polygon').iterrows():
    print(style['ESRI_STYLE'] + " : " + style['MARKER'])
                           cmap = 'RdPu',  # use a red to purple color map


The Spatially Enabled DataFrame gives you powerful visualization capabilities that allows you to plot your data on the interactive map widget. You specify colors and symbols using the same syntax that you would specify for a normal Pandas or a matplotlib plot.

Feedback on this topic?