QGIS Data Visualization

This plugin consists of a D3 circular histogram heatmap using date, time, and custom categories in the data. It also has an experimental word cloud generator, but requires adding additional python libraries and uninstalling and reinstalling the pillow python library.

QGIS D3 Date and Time Heatmap

This creates a D3 circular histogram heatmap using date, time, and custom categories in the data. An optional legend can be included.

The plugin works by counting the number of date/time/category events using two axis and displays the results as a circular heatmap. It allows you to analyze the temporal distribution of data and how much of it exists across time, based on two frequencies.

QGIS Layer Tab

This tab is for choosing the vector layer that will be used for generating the heatmap chart. Additionally, date and time fields are selected. A field representing a category can be specified and the axis of the resulting chart is configured.

• Input Vector Layer - This is a list of all the vector layers in the QGIS project. Select the one that you want to work with.
• Date/Time Field - By checking the radio button next to this filed you can specify a field that contains a date/time column. This can either be a QDateTime column or a string representing a date and time. Date/time strings are parsed using the Python dateutil dateutil.parser.parse parser. The format of the string must match one of the supported formats used by dateutil.
• Date Field - If the vector layer has separate columns for Date and Time then select this radio button and then select a field that contains a date.
• Time Field - This is enabled by checking the radio button next to the Date Field. Here you can choose a field that contains time.
• Radial Lines and Concentric Bands represent one of the following for their respective axis:
  • Year
  • Month
  • Day of Month
  • Day of Week
  • Hour of Day
  • Custom Category Field - Select this if you want to use one of the columns as a category along one of the axis.
• Custom Category specifies one of the columns that represents a category that can be selected for the axis. Both Radial Lines and Concentric Bands can select this field. If there are too many unique values in the category column, then an error message will be generated. You can limit the number unique category values by setting a Filter on the layer. See the example below.

Note that it is not necessary to have both date and time information. In this case, select either the radial box representing date and time in the same column or separate date and time columns.

Titles Tab

The chart title, legend title, and the mouse over prefix can be configured on this tab.

• Chart Titles - This text will be displayed as a header at the top of the page.
• Legend Title - If a legend is displayed, this is the text that will be displayed above it.
• Mouse Over Prefix - When Show mouse over histogram value is selected, whenever the mouse is over a part of the chart, the value of the pie segment will be displayed. This is the prefix that accompanies the pie segment value.
• Show Legend - Check this box to display the legend.
• Show mouse over histogram value - Check this to display the histogram value when the mouse is over a wedge of the heat map chart.

Settings Tab

Chart and legend settings for the heat map can be configured on this tab.

Chart Settings

• Inner radius - The radius in pixels where the first band of the heatmap starts.
• Band height - Height in pixels of each band.
• Show radial labels - When checked, the radial labels will be displayed.
• Show band labels - When checked, the labels for each band will be displayed.

Legend Settings

The legend will be displayed to the right of the chart and these are its settings.

• Maximum Legend Height - Specifies the maximum height of the legend in pixels.
• Legend Width - Specifies the width of the legend bar in pixels.
• Legend Box Width - Specifies the overall D3 legend container width. This must be large enough to include the legend bar and labels.

Colors Tab

These are settings for fine control over the heatmap color ramp and No Data Color

Tutorial Examples

These examples make use of the 2006 Chicago crime data. In 2006 there were 367,968 crime incidents. That is a lot of crime during a year in just one city. Although this author is not trained in crime, some basic conclusions can be obtained by looking at the crime incidents based on the Month of the Year, Time of the Day, and Day of the Week. Here are the examples.

Month of the year vs. day of the week.

By looking at this graph one can make some observations. The least amount of crime occurs during the months of January, February and March. This is probably due to the cold. If that is the case, why does December have a high crime rate? That is probably because of the holiday season.

Hour of the day vs. day of the week.

In this graph it appears that crime increases from 6pm to 11pm. The worst crime occurs from midnight to 1am. More difficult jobs must be taking place on the weekend where there is an increase of activity from 1am to 3am. During the day from Noon to 1pm there is also an increase of crime.

Hour of the day vs. month of the year

Here one can see a similar pattern. Crime is less during January, February, and March. Crime increases starting from about 6pm to 11pm and then from Midnight to 1am plans are executed. This graph also shows the increased crime from Noon to 1pm.

Hour of the day vs custom crime category

In this example the input data has been filtered by setting a QGIS filter for only certain types of crime. For this Example, right-mouse click on the Chicago crime layer in QGIS and select Filter then use the following filter:

"Primary Ty" IN ('THEFT','ROBBERY','OTHER OFFENSE','NARCOTICS',
    'MOTOR VEHICLE THEFT','DECEPTIVE PRACTICE','CRIMINAL TRESPASS',
    'BURGLARY','BATTERY','ASSAULT')

Set the following to:

• Radial Lines - Hour of Day
• Concentric Bands - Custom Category Field
• Custom Category - Primary Ty

This shows the resulting chart that is created.

Word Cloud Algorithms

These are experimental algorithms that generate word cloud images. They will not be available by default because the python worldcloud library needs to be installed. This is done by opening the OSGeo4W Shell and then running the command pip install wordcloud. You can then launch QGIS and the alogirhtms Word cloud from attribute and Word cloud from file will be added to the menu. If you get an error message when you run these algorithms that says “ImportError: The _imagingft C module is not installed”, you will need to run the command pip uninstall pillow and then pip install pillow. There seems to be some problem with the pillow library in QGIS. This requires system administrator privileges. On Windows find the OSGeo4W Shell in the Start menu. Right-mouse click on OSGeo4W Shell, click on More and then click on Open file location. Find the OSGeo4W Shell in the file browser window and right-mouse click on it and select Run as administrator. This will put you in administrator mode when you uninstall and reinstall pillow.

Word cloud from attritute

This algorithm generates a word cloud from one of the string attributes of a vector or table layer. It takes all of the attribute’s string values and generates a word cloud from them. The parameters are fairly self explanatory.

This is an example of the algorithm, run on the Natural Earth airport shapefile with the airport name field used as the selected attribute. The output file can be a PNG, TIFF, or SVG format.

This is the resulting output.

Word cloud from file

This algorithm generates a word cloud from all the words in a .txt file. This shows the processing algorithm configured to run on US Constitution.

This shows the resulting word cloud output generated from the US Constitution.