refactor: use sum in sql query

and set zero_devision to nan for report

Closes #899

Author: matthiasschaub

ohsome_quality_api/indicators/land_cover_thematic_accuracy/indicator.py

@@ -3,6 +3,7 @@
 from pathlib import Path
 from string import Template
 
+import numpy
 import plotly.graph_objects as pgo
 from geojson import Feature
 from sklearn.metrics import classification_report, confusion_matrix, f1_score
@@ -137,15 +138,12 @@ def calculate(self) -> None:
             f"{label_description} {self.templates.result_description}"
         )
 
-        # TODO: UdefinedMetricWarning
-        # Recall is ill-defined and being set to 0.0 in labels with no
-        # true samples. Use `zero_division` parameter to control this
-        # behavior.
         # NOTE: For introspection/testing only
         self.report = classification_report(
             self.clc_classes_corine,
             self.clc_classes_osm,
             sample_weight=self.areas,
+            zero_division=numpy.nan,
         )
 
     def create_figure(self) -> None:

ohsome_quality_api/indicators/land_cover_thematic_accuracy/query-multi-classes.sql

@@ -6,13 +6,18 @@ WITH bpoly AS (
 SELECT
     CLC_class as clc_class_corine,
     osm_CLC_class as clc_class_osm,
-    CASE WHEN ST_Within (o.geometry, b.geometry) THEN
-        area
-    ELSE
-        ST_Area (ST_Intersection (o.geometry, b.geometry)::geography)
-    END AS area
+    SUM (
+        CASE WHEN ST_Within (o.geometry, b.geometry) THEN
+            area
+        ELSE
+            ST_Area (ST_Intersection (o.geometry, b.geometry)::geography)
+        END) AS area
 FROM
     osm_corine_2021_deu_intersection o,
     bpoly b
 WHERE
-    ST_Intersects (o.geometry, b.geometry);
+    ST_Intersects (o.geometry, b.geometry)
+    --AND osm_CLC_class != 0
+    --AND osm_CLC_class != 50
+    GROUP BY 1,2
+;

tests/approvals/integrationtests/indicators/test_land_cover_thematic_accuracy.py-test_calculate_multi_class/report.approved.txt

@@ -1,19 +1,19 @@
               precision    recall  f1-score   support
 
-           0       0.00      0.00      0.00       0.0
+           0       0.00       nan      0.00       0.0
           11       0.94      0.86      0.90 13.681696054340627
           12       0.84      0.87      0.85 4.868706606602494
-          13       0.00      0.00      0.00       0.0
+          13       0.00       nan      0.00       0.0
           14       0.13      0.42      0.20 0.3617314972878062
           21       0.86      0.91      0.89 15.526750634345152
           22       0.64      0.76      0.70 2.2997486561073313
           23       0.75      0.44      0.55 6.712711587504546
-          24       0.00      0.00      0.00 0.084999664294624
+          24        nan      0.00      0.00 0.084999664294624
           31       0.82      0.92      0.87 10.490463269571688
           32       0.00      0.00      0.00 0.2576135038588164
-          50       0.00      0.00      0.00       0.0
-          51       0.00      0.00      0.00 0.033118424422509736
+          50       0.00       nan      0.00       0.0
+          51        nan      0.00      0.00 0.033118424422509736
 
     accuracy                           0.82 54.31753989833559
-   macro avg       0.38      0.40      0.38 54.31753989833559
+   macro avg       0.45      0.52      0.38 54.31753989833559
 weighted avg       0.84      0.82      0.82 54.31753989833559