#20

Created co-occurence matrix - file based per found element in corpus

get_compounds.py

@@ -9,6 +9,7 @@ from tqdm import tqdm
 from collections import Counter
 from nltk.tokenize import word_tokenize
 nltk.download('punkt')
+import csv
 
 
 class Document:
@@ -67,7 +68,19 @@ if __name__ == '__main__':
     # calculate total count of element
     result['total_count'] = result['count_per_file'].apply(sum)
     result = result.sort_values('total_count')
-    print(result)
+    #print(result)
     result.to_csv("compounds_found.csv", index=False)
+    co_occ = result.drop('count_per_file', axis=1)
+    co_occ = co_occ.drop('total_count', axis=1)
+    # create dict for co-occurence matrix
+    co_occ_dict = dict(zip(co_occ.element, co_occ.file))
+    # create co-occurence matrix
+    values = sorted(set(e for v in co_occ_dict.values() for e in v))
+    co_occ_matrix = {k: [1 if value in v else 0 for value in values] for k, v in co_occ_dict.items()}
+    with open('co_occurence.csv', 'w') as f:
+        writer = csv.writer(f)
+        writer.writerow(values)
+        for k, v in co_occ_matrix.items():
+            writer.writerow([k, v])