A Quantum GIS plug-in to apply Bayesian belief network models on raster data
© 2015, Dries Landuyt (mailto:drieslanduyt@gmail.com)
Step 1. Save the PMAT folder, present in this repository, to the following directory of Quantum GIS:
...\apps\qgis\python\plugins\
Step 2. Copy the Netica.dll
file which can be found in your local installation of Netica or at the Norsys website to the following directory of your Quantum GIS installation:
...\bin\
Whether you need the 32bit or 64bit version of the Netica.dll
will depend on the python installation that is used by Quantum GIS. To check this, run the following code in the python console of QGIS:
import struct
print 8*struct.calcsize("P")
Step 1. Develop a Bayesian belief network model by using the graphical user interface of Netica
Specific model requirements for the plugin include:
OUT
and IN
IN
nodeset should have statenames assigned to each of their statesOUT
nodeset should have numerical statetitles assigned to each of their states Step 2. Prepare input maps
Convert the input maps to GeoTiff format and name them according to the names of the corresponding input nodes' names (nodename.tif
). Store these files (including the legend files (see below)) in the same folder as the one your network file is in. Prepare for each GeoTiff input map legend files. A legend file assigns a statename to each numerical value (not for values that represent no data) included in the GeoTiff file. The legend file should be named and structured as follows:
nodenameleg.csv
numerical value 1 statename 1
numerical value 2 statename 2
numerical value 3 statename 3
numerical value 4 statename 4
numerical value 5 statename 5
. .
. .
. .
numerical value n statename n
Examples of correctly formatted GeoTiff files, legend files and network file can be found in the Example folder
Step 3. Open Quantum GIS
via the Plugins
tab you should be able to open the plug-in. In case the plug-in does not appear in the list, search for it via Plugins
-> Manage and Install Plugins...
Step 4. Browse to the proper directory
This directory should contain the input maps (.tif), the legend files (.csv) and the network file (.neta). To test the plugin, use the Example folder provided in this repository.
Step 5. Browse to the proper network file (.neta)
To test the plugin, select the Example_Network.neta file
Step 6. Select the type(s) of output map(s) you want
Step 7. Select a calculation method
Fast
: The network model will be transformed into a look-up table. This look-up table will be used to link each pixel with it's probabilistic model output. Slow
: The network model will be ran on each pixel seperately. Step 8. Specify whether you want the output being visualised on the map canvas
Select or deselect the Add map to canvas
checkbox
Step 9. Click the OK
button, the progress bars will load successively
For questions or bug reports, please open an issue in this repository