r.confusionmatrix: Support usage of a single class in reference (#409)

* add handling for single class
* remove unnecessary line printing
* add test
* change csv.writer linedelimiter to `\n`

grass7/raster/r.confusionmatrix/r.confusionmatrix.py

@@ -178,7 +178,6 @@ def get_r_kappa(classification, reference):
     grass.run_command(
         'r.kappa', flags='wmh', classification=classification,
         reference=reference, output=tmp_csv, quiet=True)
-
     # read csv: error matrix
     errorlist = None
     with open(tmp_csv) as csvfile:
@@ -220,9 +219,12 @@ def convert_output(classified_classes, ref_classes, confusionmatrix, overall_acc
     line2.extend(confusionmatrix[0,:].tolist()[0])
     line2.extend([user_accuracy[classified_classes[0]], commission[classified_classes[0]]])
 
-    line3 = [classification, classified_classes[1]]
-    line3.extend(confusionmatrix[1,:].tolist()[0])
-    line3.extend([user_accuracy[classified_classes[1]], commission[classified_classes[1]]])
+    if len(ref_classes) == 1:
+        line3 = [classification, '', '', '', '']
+    else:
+        line3 = [classification, classified_classes[1]]
+        line3.extend(confusionmatrix[1,:].tolist()[0])
+        line3.extend([user_accuracy[classified_classes[1]], commission[classified_classes[1]]])
 
     lines = [["", "", "Reference Map", refname], line1, line2, line3]
     for i in range(2, confusionmatrix.shape[0]):
@@ -231,7 +233,7 @@ def convert_output(classified_classes, ref_classes, confusionmatrix, overall_acc
         linei.extend([user_accuracy[classified_classes[i]], commission[classified_classes[i]]])
         lines.append(linei)
     producer_accuracy_list = [producer_accuracy[rc] for rc in ref_classes]
-    commission_list = [commission[rc] for rc in classified_classes]
+    # commission_list = [commission[rc] for rc in classified_classes]
     lineend1 = ["", "Producer Accuracy"]
     lineend1.extend(producer_accuracy_list)
     lineend1.extend(['Overall Accuracy', overall_accuracy])
@@ -339,6 +341,17 @@ def main():
     if not flags['m']:
         grass.message("\nKappa coefficient: %f" % kappa)
 
+    # round values to two digits
+    for item in user_accuracy.items():
+        user_accuracy[item[0]] = round(item[1], 2)
+    for item in producer_accuracy.items():
+        producer_accuracy[item[0]] = round(item[1], 2)
+    for item in commission.items():
+        commission[item[0]] = round(item[1], 2)
+    for item in omission.items():
+        omission[item[0]] = round(item[1], 2)
+    overall_accuracy = round(overall_accuracy, 2)
+    kappa = round(kappa, 2)
     # in matrix style
     if flags['m'] or options['csvfile']:
         lines = convert_output(
@@ -349,14 +362,16 @@ def main():
     # write csv file
     if csv_filename:
         with open(csv_filename, 'w') as file:
-            writer = csv.writer(file)
+            writer = csv.writer(file, lineterminator='\n')
             for line in lines:
                 writer.writerow(line)
     if flags['m']:
         for line in lines:
-            # for stdout using print
             print(line)
 
+    if len(ref_classes) == 1:
+        grass.warning(_('Only one class in reference dataset.'))
+
     # cleanup
     if options['vector_reference']:
         grass.run_command('g.remove', type='raster', name=reference, flags='f', quiet=True)

grass7/raster/r.confusionmatrix/testsuite/data/confusionmatrix_raster_matrix.csv

@@ -1,19 +1,19 @@
-,,Reference Map,landclass96
-,,developed,agriculture,herbaceous,shrubland,forest,water,sediment,Class_7,Class_8,Class_9,Class_10,Class_11,Class_12,Class_13,Class_14,User Accuracy,Commission Error
-Classified Map,High Intensity Developed,34931,0,0,0,0,0,0,0,0,0,0,0,0,0,0,100.0,0.0
-landuse96_28m,Low Intensity Developed,38146,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
-,Cultivated,0,2137,0,0,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
-,Managed Herbaceous Cover,0,0,25488,0,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
-,Upland Herbaceous,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
-,Riverine/Estuarine Herbaceous,0,0,30,0,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
-,Evergreen Shrubland,0,0,0,16464,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
-,Deciduous Shrubland,0,0,0,320,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
-,Mixed Shrubland,0,0,0,45,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
-,Mixed Hardwoods,0,0,0,0,7893,0,0,0,0,0,0,0,0,0,0,0.0,100.0
-,Bottomland Hardwoods/Hardwood Swamps,0,0,0,0,19733,0,0,0,0,0,0,0,0,0,0,0.0,100.0
-,Southern Yellow Pine,0,0,0,0,64774,0,0,0,0,0,0,0,0,0,0,0.0,100.0
-,Mixed Hardwoods/Conifers,0,0,0,0,33865,0,0,0,0,0,0,0,0,0,0,0.0,100.0
-,Water Bodies,0,0,0,0,0,5303,0,0,0,0,0,0,0,0,0,0.0,100.0
-,Unconsolidated Sediment,0,0,0,0,0,0,194,0,0,0,0,0,0,0,0,0.0,100.0
-,Producer Accuracy,47.800265473404764,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,Overall Accuracy,14.010340000721955
-,Omission Error,52.199734526595236,100.0,100.0,100.0,100.0,100.0,100.0,100.0,100.0,100.0,100.0,100.0,100.0,100.0,100.0,Kappa coefficient,0.09464798548745997
+,,Reference Map,landclass96
+,,developed,agriculture,herbaceous,shrubland,forest,water,sediment,Class_7,Class_8,Class_9,Class_10,Class_11,Class_12,Class_13,Class_14,User Accuracy,Commission Error
+Classified Map,High Intensity Developed,34931,0,0,0,0,0,0,0,0,0,0,0,0,0,0,100.0,0.0
+landuse96_28m,Low Intensity Developed,38146,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
+,Cultivated,0,2137,0,0,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
+,Managed Herbaceous Cover,0,0,25488,0,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
+,Upland Herbaceous,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
+,Riverine/Estuarine Herbaceous,0,0,30,0,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
+,Evergreen Shrubland,0,0,0,16464,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
+,Deciduous Shrubland,0,0,0,320,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
+,Mixed Shrubland,0,0,0,45,0,0,0,0,0,0,0,0,0,0,0,0.0,100.0
+,Mixed Hardwoods,0,0,0,0,7893,0,0,0,0,0,0,0,0,0,0,0.0,100.0
+,Bottomland Hardwoods/Hardwood Swamps,0,0,0,0,19733,0,0,0,0,0,0,0,0,0,0,0.0,100.0
+,Southern Yellow Pine,0,0,0,0,64774,0,0,0,0,0,0,0,0,0,0,0.0,100.0
+,Mixed Hardwoods/Conifers,0,0,0,0,33865,0,0,0,0,0,0,0,0,0,0,0.0,100.0
+,Water Bodies,0,0,0,0,0,5303,0,0,0,0,0,0,0,0,0,0.0,100.0
+,Unconsolidated Sediment,0,0,0,0,0,0,194,0,0,0,0,0,0,0,0,0.0,100.0
+,Producer Accuracy,47.8,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,Overall Accuracy,14.01
+,Omission Error,52.2,100.0,100.0,100.0,100.0,100.0,100.0,100.0,100.0,100.0,100.0,100.0,100.0,100.0,100.0,Kappa coefficient,0.09

grass7/raster/r.confusionmatrix/testsuite/data/confusionmatrix_raster_matrix_oneclass.csv

@@ -0,0 +1,6 @@
+,,Reference Map,landclass96_class1only
+,,Class_0,User Accuracy,Commission Error
+Classified Map,Class_0,73077,100.0,0.0
+landuse96_28m_class1only,,,,
+,Producer Accuracy,100.0,Overall Accuracy,100.0
+,Omission Error,0.0,Kappa coefficient,nan

grass7/raster/r.confusionmatrix/testsuite/test_r_confusionmatrix.py

@@ -13,10 +13,13 @@
 
 from grass.gunittest.case import TestCase
 from grass.gunittest.main import test
+from grass.gunittest.gmodules import SimpleModule
 
 class Testconfusionmatrix(TestCase):
     classification = 'landuse96_28m'
+    classification_oneclass = 'landuse96_28m_class1only'
     raster_ref = 'landclass96'
+    raster_ref_oneclass = 'landclass96_class1only'
     output_csv = 'output_confusionmatrix.csv'
 
     @classmethod
@@ -25,11 +28,19 @@ def setUpClass(cls):
         cls.use_temp_region()
         # set region to classification
         cls.runModule('g.region', raster=cls.classification)
+        cls.runModule('r.mapcalc', expression='%s = if(%s==1,1, null())' % (
+         cls.raster_ref_oneclass, cls.raster_ref))
+        cls.runModule('r.mapcalc', expression='%s = 1' % (
+         cls.classification_oneclass))
 
     @classmethod
     def tearDownClass(cls):
         """Remove the temporary region and generated data"""
         cls.del_temp_region()
+        cls.runModule('g.remove', type='raster',
+                      name='%s,%s' % (
+                       cls.raster_ref_oneclass, cls.classification_oneclass),
+                      flags='f')
 
     def tearDown(self):
         """Remove the outputs created
@@ -50,10 +61,29 @@ def test_confusionmatrix_with_raster_reference(self):
         self.assertFileExists(self.output_csv,
                                msg='Output file does not exist')
         # check if the output file is equal to the reference file
+
         self.assertFilesEqualMd5(self.output_csv,
                                'data/confusionmatrix_raster_matrix.csv',
                                msg='Output file is not equal to reference file')
 
+    def test_confusionmatrix_with_singleclass(self):
+        """ Test confusionmatrix with a single class reference and classification"""
+        r_confusion = SimpleModule('r.confusionmatrix',
+                      classification=self.classification_oneclass,
+                      raster_reference=self.raster_ref_oneclass,
+                      flags='m',
+                      csvfile=self.output_csv)
+        self.assertModule(r_confusion)
+        # check to see if output file exists
+        self.assertFileExists(self.output_csv,
+                               msg='Output file does not exist')
+        # check if the output file is equal to the reference file
+        self.assertFilesEqualMd5(self.output_csv,
+                               'data/confusionmatrix_raster_matrix_oneclass.csv',
+                               msg='Output file is not equal to reference file')
+        #test that the warning is shown
+        stderr = r_confusion.outputs.stderr
+        self.assertIn('Only one class in reference dataset.', stderr)
 
 
 if __name__ == '__main__':