Add a script and a rule to generate the centralities histogram from the actual...

Add a script and a rule to generate the centralities histogram from the actual predictions of the model

Makefile

@@ -106,6 +106,9 @@ data/corpus/domainGroup_frequencies.tsv: data/corpus/metadata.tsv
 %_frequencies.tsv: $$(subst _frequencies,,$$(*)).tsv
 	./visualisation/SortedFrequencies.py $< $@ domainGroup
 
+%/centralities.tsv: %/confusionMatrix.json
+	./visualisation/EigenvectorCentrality.py $< $@
+
 figure/histogram/%.png: data/%.tsv
 	./visualisation/BarPlot.py $< $@
 

visualisation/EigenvectorCentrality.py

+#!/usr/bin/env python3
+
+from GEODE import toTSV
+from GEODE.Visualisation.ConfusionMatrix import getConfusionMatrix
+from IterateMatrix import iterate
+import pandas
+import sys
+
+def eigenVector(inputJSON, outputTSV):
+    longEnough = 1000
+    m = getConfusionMatrix(inputJSON)
+    centralities = pandas.Series(data=iterate(m, longEnough)['matrix'][0],
+                                 index=m['labels'],
+                                 name='Centralité')
+    centralities.sort_values(ascending=False, inplace=True)
+    centralities.index.name = 'domainGroup'
+    centralities.to_csv(outputTSV, sep='\t')
+
+if __name__ == '__main__':
+    eigenVector(*sys.argv[1:])