Section: Misc. Reference Manual Pages (04 September 2013)
om_test - test a distribution model using the openModeller framework
om_test [-] v --version | r --xml-req FILE | o --model FILE p --points FILE [ --calc-matrix t --threshold VALUE ] [ --calc-roc --num-background VALUE --max-omission VALUE ] [ s --result FILE ] [ --log-level LEVEL ] [ --log-file FILE ] [ --prog-file FILE ]
om_test is a command line tool to test distribution models. There are two ways of providing input: one is to specify an XML file containing a test request according to the TestParameters element definition in the openModeller XML Schema located in http://openmodeller.cria.org.br/xml/1.0/openModeller.xsd (see also test_request.xml in the openModeller examples directory) and the other is to specify a file with a serialized model according to the SerializedModel element definition in the openModeller XML Schema together with another file with the points to be tested (TAB-delimited, following the same pattern used to specify points in text files for om_console). When providing a serialized model and a file with points to be tested, these points must have the same spatial reference and label as the training points that can be found in the serialized model. In this case, the layers used during the test will also be the same ones used to create the model. Test results include confusion matrix and ROC curve. Results will be either displayed on the screen or stored in another file if the corresponding parameter was specified. The result will be in XML, following the ModelStatisticsType definition in the openModeller XML Schema. Serialized models can be generated with om_console or om_model.
- -v, --version
Display version info.
- -r, --xml-req
File containing a test request in XML.
- -o, --model
File containing a serialized model in XML.
- -p, --points
File containing the points to be tested.
Calculate confusion matrix for the training data.
- -t --threshold
Probability threshold to distinguish between presence/absence when calculating the confusion matrix.
Calculate ROC curve for the training data.
Number of background points to generate the ROC curve when there are no absences.
Calculate partial area ratio for points under the maximum omission.
- -s, --result
File where the test result will be stored.
openModeller log level: debug, warn, info or error. Defaults to "info".
File to store log information.
File to store progress (-1=queued, -2=aborted, -3=cancelled, [0,100]=progress).
-c, --config-file Configuration file for openModeller (available since version 1.4).
Renato De Giovanni <renato at cria dot org dot br>
This document was created by
using the manual pages.
Time: 22:44:24 GMT, April 28, 2014