Update clustering.py

clustering.py

@@ -706,18 +706,19 @@ def build_sankey_data(
     values = []
     comments = []
 
-    threshold_values = sorted(cluster_assignments.keys())
-    log_message(f"build_sankey_data {threshold_values}")
+    sorted_threshold_values = sorted(cluster_assignments.keys())
+    log_message(f"build_sankey_data {sorted_threshold_values}")
     valid_clusters = {}
 
-    for threshold in threshold_values:
+    for threshold in sorted_threshold_values:
+        print(threshold)
         valid_clusters[threshold] = [
             j
             for j in np.unique(cluster_assignments[threshold])
             if cluster_counts[threshold].get(j, 0) >= min_items_by_cluster
         ]
 
-    for i, threshold in enumerate(threshold_values):
+    for i, threshold in enumerate(sorted_threshold_values):
         for j in valid_clusters[threshold]:
             cluster_name = (
                 f"{j} (d={threshold})\nTotal: {cluster_counts[threshold].get(j, 0)}"
@@ -733,7 +734,7 @@ def build_sankey_data(
                 )
 
         if i > 0:
-            prev_threshold = threshold_values[i - 1]
+            prev_threshold = sorted_threshold_values[i - 1]
             for prev_cluster in valid_clusters[prev_threshold]:
                 for curr_cluster in valid_clusters[threshold]:
                     count = np.sum(