mdod 1.0.9__tar.gz → 2.0.0__tar.gz

{mdod-1.0.9 → mdod-2.0.0}/PKG-INFO

@@ -1,11 +1,12 @@
 Metadata-Version: 2.1
 Name: mdod
-Version: 1.0.9
+Version: 2.0.0
 Summary: MDOD, Multi-Dimensional data Outlier Detection
 Home-page: https://github.com/mddod/mdod
 Author: Z Shen
 Author-email: 626456708@qq.com
 License: BSD 3-Clause License
+Platform: UNKNOWN
 Classifier: Development Status :: 5 - Production/Stable
 Classifier: Intended Audience :: Developers
 Classifier: Programming Language :: Python
@@ -82,3 +83,5 @@ data1,data2,data3,data4,data5,data6
 Please visit https://github.com/mddod/mdod, or https://mddod.github.io/
 
  
+
+

mdod-2.0.0/mdod/mdod.py

@@ -0,0 +1,55 @@
+import numpy as np
+
+def md(detections, norm_distance, top_n):
+    """
+    Calculate the vector cosine similarity score and return the sum of the first top_n scores of each vector.
+    
+    Args:
+        detections: Array data
+        norm_distance: Standardized distance value (scalar) in the new dimension
+        top_n: The number of highest similarity scores retained
+    
+    Returns:
+        List of [score_sum, original_vector, index]
+    """
+    # Input verification: check whether detections is empty or whether top_n is valid
+    if detections is None or detections.size == 0 or top_n <= 0:
+        return []
+    
+    # Make sure the input is a NumPy array
+    dets_array = np.array(detections, dtype=float)
+    n_samples, n_features = dets_array.shape
+    
+    # Precomputed constant term
+    nd_squared = norm_distance ** 2  
+    denominator_left = norm_distance  
+    
+    # Result storage
+    result_list = []
+    
+    # Vectorize each vector
+    for i in range(n_samples):
+        current_vector = dets_array[i]
+        
+        # Calculate the sum of squares of the difference between all vectors and the current vector
+        diff = dets_array - current_vector  
+        diff_squared_sum = np.sum(diff ** 2, axis=1)  
+        
+        # Calculate the right part of the denominator
+        denominator_right = np.sqrt(diff_squared_sum + nd_squared)  
+        denominator = denominator_left * denominator_right  
+        
+        # Calculate the molecule
+        numerator = nd_squared * np.ones(n_samples)  
+        
+        # Calculate the similarity score and avoid dividing by zero
+        similarity_scores = np.where(denominator == 0, 0, numerator / denominator)
+        similarity_scores[i] = -np.inf  
+        
+        # Get the sum of the top_n scores
+        top_scores_sum = np.sum(np.partition(similarity_scores, -top_n)[-top_n:])
+        
+        # Save result
+        result_list.append([top_scores_sum, detections[i].tolist(), i])
+    
+    return result_list

{mdod-1.0.9 → mdod-2.0.0}/mdod.egg-info/PKG-INFO

@@ -1,11 +1,12 @@
 Metadata-Version: 2.1
 Name: mdod
-Version: 1.0.9
+Version: 2.0.0
 Summary: MDOD, Multi-Dimensional data Outlier Detection
 Home-page: https://github.com/mddod/mdod
 Author: Z Shen
 Author-email: 626456708@qq.com
 License: BSD 3-Clause License
+Platform: UNKNOWN
 Classifier: Development Status :: 5 - Production/Stable
 Classifier: Intended Audience :: Developers
 Classifier: Programming Language :: Python
@@ -82,3 +83,5 @@ data1,data2,data3,data4,data5,data6
 Please visit https://github.com/mddod/mdod, or https://mddod.github.io/
 
  
+
+

{mdod-1.0.9 → mdod-2.0.0}/setup.py

@@ -5,7 +5,7 @@ with open("README.md", "r", encoding="utf-8") as fh:
 
 setup(
     name="mdod",
-    version="1.0.9",
+    version="2.0.0",
     packages=find_packages(),
     package_data={"": ["*"]},  
     install_requires=[],

mdod-1.0.9/mdod/mdod.py

@@ -1,39 +0,0 @@
-# -*-coding: utf-8- -*-
-#MDOD, Multi-Dimensional data Outlier Detection
-# Author: Z Shen<626456708@qq.com>
-# License: BSD 3-Clause License
-
-import numpy as np
-
-def md(dets0, nd, sn):
-    VCS_list = []
-    i = 0
-
-    for line0 in dets0:
-        VCSResult_list = []
-        line0_arr = np.array(line0, dtype=float)  
-
-        for j, line1 in enumerate(dets0):
-            if j == i:  
-                continue
-
-            line1_arr = np.array(line1, dtype=float)
-
-            DenominatorLeft = np.sqrt(np.sum((line0_arr - line0_arr) ** 2) + (nd - 0) ** 2)
-            DenominatorRight = np.sqrt(np.sum((line0_arr - line1_arr) ** 2) + (nd - 0) ** 2)
-            DenominatorSum = DenominatorLeft * DenominatorRight
-
-            NumeratorSum = np.sum(np.sqrt((line0_arr - line0_arr) ** 2) * np.sqrt((line0_arr - line1_arr) ** 2))
-            NumeratorPlus = np.sqrt((nd - 0) ** 2) * np.sqrt((nd - 0) ** 2)
-            NumeratorSum += NumeratorPlus
-
-            VCSResult = 0 if DenominatorSum == 0 else NumeratorSum / DenominatorSum
-            VCSResult_list.append(VCSResult)
-
-        VCSResult_list.sort(reverse=True)
-        VCSTotal = sum(VCSResult_list[:min(sn, len(VCSResult_list))]) 
-
-        VCS_list.append([VCSTotal, line0.tolist(), i])  
-        i += 1
-
-    return VCS_list