pytrendy 1.2.0.dev3__tar.gz → 1.2.0.dev4__tar.gz

{pytrendy-1.2.0.dev3 → pytrendy-1.2.0.dev4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pytrendy
-Version: 1.2.0.dev3
+Version: 1.2.0.dev4
 Summary: Trend Detection in Python. Applicable for real-world industry use cases in time series.
 License: MIT License
          

{pytrendy-1.2.0.dev3 → pytrendy-1.2.0.dev4}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "pytrendy"
-version = "1.2.0.dev3"
+version = "1.2.0.dev4"
 description = "Trend Detection in Python. Applicable for real-world industry use cases in time series."
 authors = [
     { name = "Russell Sammut Bonnici", email = "r.sammutbonnici@gmail.com" },

{pytrendy-1.2.0.dev3 → pytrendy-1.2.0.dev4}/pytrendy/post_processing/segments_refine/abrupt_shaving.py

@@ -123,7 +123,7 @@ def shave_abrupt_trends(df: pd.DataFrame, value_col: str, segments: list[dict],
 
     # Second pass to pad segments if specified
     segments_padded = deepcopy(segments_refined)
-    if method_params.get('abrupt_padding', 0) >= 0:
+    if method_params.get('abrupt_padding', 0) > 0:
 
         meta_df = pd.DataFrame(segments_refined) # metadata df, to filter by datetime easily
         meta_df['start'] = pd.to_datetime(meta_df['start'])

{pytrendy-1.2.0.dev3 → pytrendy-1.2.0.dev4}/pytrendy/post_processing/segments_refine/trend_classify.py

@@ -36,6 +36,11 @@ def classify_trends(df: pd.DataFrame, value_col: str, segments: list[dict]) -> l
         if segment['direction'] not in ['Up', 'Down']: 
             continue
 
+        # Skip reclassification for padded segments — their length is artificially inflated
+        # by abrupt_padding, so DTW would wrongly classify them as gradual
+        if segment.get('padded', False):
+            continue
+
         # Assume some padding for abrupt cases
         start = pd.to_datetime(segment['start']) - pd.Timedelta(days=2)
         end = pd.to_datetime(segment['end']) + pd.Timedelta(days=2)