Combining and Analyzing Data in QGIS

Link to slides: https://docs.google.com/presentation/d/19zR6xYUu7FWxby94WDf7TaAuxhOwnoAo-TAVW17DZwE/edit

Analyze Grocery Store Data

Load in Shapefiles and Text Data

Load Shapefiles

In QGIS, go to Layer > Add Layer > Add Vector Layer
In the pop-up window that opens:
- Source Type = File should be the default, but if it isn’t, select that.
- In Vector Dataset(s) under 'Source', open the file finder (3 dots to the right)
  - In the data provided, select Georgia_Census_Tracts > Georgia_Census_Tracts.shp.
- Once the .shp file is selected, click ‘Add’.
Repeat this process with the Grocery_Stores_Atlanta_Metro shapefile.
Click ‘Close’ to close the window and view the data. You should see both datasets load in the view and their names listed in the 'Layers' section to the bottom left.

Note: You should be able to load shapefile data by selecting any of the main 3 files in the folder (.shp, .dbf, .shx). Explanations of file types are below for those interested!

.shp: contains the geometry for all features.
.shx: indexes the geometry.
.dbf: stores feature attributes in a table.

These should be in every shapefile, you may also see:

.prj: contains information on projection format including the coordinate system and projection information
.sbn and .sbx: spatial index of the features.
.xml: geospatial metadata in XML format

Load Text Data

Go to Layer > Add Layer > Add Delimited Text Layer
Browse to the Income folder in the data and select the .csv in the folder named 'income.csv' and click ‘Open’
Under Geometry Definition, make sure ‘No geometry’ is selected. (Census data will have GEOIDs we can use to join with geometric data, but it does not contain geometry itself.)
Repeat the process for ‘Race data’

Congratulations! You should now have four layers in the QGIS layers panel.

Join Data

Filter Shapefile by ID to Get Counties of Interest

For the Georgia_Census_Tracts, go to Attribute Table and click on 'Select Features by Expression" (looks like an E and a square)
In the expression area, paste the formula below

array_contains( array('13057', '13063', '13067', '13089', '13097', '13113', '13121', '13135', '13151', '13247'), left( "GEOID", 5))
Click 'Select Features' on the bottom right, and then hit 'Close'
- You should see the tracts within those counties highlighted
Right click on the file and go to Export > Export Selected Features As
- Important: select the 'Selected' Features option, which is the second option, not just export features as
And save files as Atlanta_Metro_Census_Tracts or something similar. That should now appear in your layers bar.
Unselect or Remove Georgia_Census_Tracts, we won't be using it anymore.

Change field values in QGIS

To join data with geometries to data without geometries, we need a field to join across datasets.

To inspect a dataset's attributes, right click on the dataset in the Layers panel (bottom left) > Open Attribute Table

Open the attribute table for either census dataset (Income or Race) and look at the column GEO_ID. It should be in the format 1400000US13001950100. Notice how they all have the prefix ‘1400000US’ — the only meaningful digits for us are the last 11: The first two digits, 13, represents the FIPS code for Georgia, the next 3 digits indicate the county, and the last 6 indicate census tracts/block etc.

Open the attribute table for the Atlanta_Metro_Census_Tracts shapefile, and you’ll see the GEOIDs are in the format 13001950100 (missing the prefix), so we need to change them to match. To do that, follow steps below:

Select the Field calculator in the top menu (looks like an abacus)
Set 'Output Field' = id or some similar name
Set 'Output Field type' = Text(string)
Set 'Output Field length' = 20 (must be large enough to accomodate a 20-letter id)
In the expression box (empty text box below), type '1400000US' + "GEOID"
- This tells the Field Calculator to create a new field of type string, but adding the 1400000US prefix to each GEOID
Click 'Ok'

Note: you can do all sort of expressions in the field calculator. The box on the right-hand side is a great starting point to explore!

Join data on Field

Double click on the Atlanta_metro_area in the Layers area
Click on the ‘Joins’ option on the left-hand side of the pop up
Click the green plus button on the bottom left
Set Join Layer = to either Income or Race data
Set Join field = GEO_ID (or whatever the id field is)
Set Target field = the new field in we created in the step above, id
Click 'Ok'. It might take a second but you should see a join appear in the Joins box

Visualize the Data (Optional)

Double click the Atlanta_Metro_Census_Tracts layer again
Go to ‘Symbology’
In 'Value', select the column you want to color by.
- If you don't see the column you want appear, it may be the wrong type. To convert from string to number, click the E sign next to the Value dropdown and paste to_int( "income_S1903_C03_001E") into the expression area.
In the dropdown at the top, select ‘Graduated’ (default should be ‘Single Symbol’)
On the bottom right, increase the number of ‘Classes’ to somewhere around 10.
Check the ranges generated are expected given whatever the data is.
Click 'Ok'

This will tell QGIS to color the geodata by whatever property you select, giving you a way to eyeball the data, looking for possible trends.

Count Number of Grocery Stores in Each Tract

Go to Vector > Analysis Tools > Count Points in Polygon
Select 'Polygons' = to the Atlanta_metro_area layer
Select the Points = to the Grocery_Stores_Atlanta_Metro layer
Make sure 'Count Field Name' = NUMPOINTS
- This tells QGIS to sum by # of points in each polygon/tract
Click ‘Run’
You should see a new layer called ‘Count’ in the layers area
Right click on ‘Count’ and go to Export > Save Features As and save the file as a .csv

You should now have a .csv with GEOID, median income, % white, population, and the # grocery stores for every tract in the Atlanta Metro Area! Analyze however you prefer — but we can talk you through some basic ideas :)

Quartile functions:

=SUM(C2:INDIRECT("C"&ROUND(738/4, 0)))

=SUM(INDIRECT("C"&ROUND(738/4, 0)+1):INDIRECT("C"&ROUND(738/4, 0) * 2))

=SUM(INDIRECT("C"&ROUND(738/4, 0)*2+1):INDIRECT("C"&ROUND(738/4, 0) * 3))

=SUM(INDIRECT("C"&ROUND(738/4, 0)*3+1):INDIRECT("C"&ROUND(738/4, 0) * 4))

Optional Next Exercise -- Area Analysis

Sometimes the geodata we want to analyse isn't lat/lngs, it's area. For instance, we might want to look at the % of each census tract in the Atlanta Metro area that is public park.

First, load in the Greenspace data set from the data folder, then follow the instructions below.

Calculate Area of Census Tracts

We need to add in an area field to each census tract, so we know the area covered by each tract

Open the Field Calculator for the Atlanta Census Tract layer
Set 'Output Field' = tract_area or some similar name
Set 'Output Field type' = Decimal number
Type the expression $area in the text field
Click 'Ok'

This might take a second, because it's calcuating the area covered by each tract.

Calculate Difference between Tract and Parks

Go to Vector > Geoprocessing Tools > Difference
Select the Input Layer = Atlanta Census Tract layer and the Overlay = Greenspace
Click 'Run'

You should see a new layer called 'Difference' outputted, unselect Greenspace and Atlanta Tract Area to see it more clearly. This contains the parts of census tracts NOT covered in park, but still contain the census tract total area as a field.

Calculate New Area of Census Tracts Without Parks

Do the same calculation we did above, just with these new polygons.

Open the Field Calculator for the Difference layer
Set 'Output Field' = tract_area or some similar name
Set 'Output Field type' = Decimal number
Type the expression $area in the text field
Click 'Ok'

Calculate the % of Total Area

Open the Field Calculator again
Set 'Output Field' = percent or some similar name
Set 'Output Field type' = Decimal number
Enter the expression below and click 'Ok'

1 - "area_part" / "area_full"

This will add a field with the percent of tract covered by greenspace.

Export and analyze!

ruthwtalbot / NICAR-2022