mlquantify 0.1.11__tar.gz → 0.1.13__tar.gz

{mlquantify-0.1.11/mlquantify.egg-info → mlquantify-0.1.13}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mlquantify
-Version: 0.1.11
+Version: 0.1.13
 Summary: Quantification Library
 Home-page: https://github.com/luizfernandolj/QuantifyML/tree/master
 Maintainer: Luiz Fernando Luth Junior

mlquantify-0.1.13/VERSION.txt

@@ -0,0 +1 @@
+0.1.13

{mlquantify-0.1.11 → mlquantify-0.1.13}/mlquantify/adjust_counting/_adjustment.py

@@ -603,7 +603,12 @@ class MS(ThresholdAdjustment):
         for thr, tpr, fpr in zip(thresholds, tprs, fprs):
             cc_predictions = CC(thr).aggregate(predictions)
             cc_predictions = cc_predictions[1]
-            prevalence = cc_predictions if tpr - fpr == 0 else (cc_predictions - fpr) / (tpr - fpr)
+            
+            if tpr - fpr == 0:
+                prevalence = cc_predictions
+            else:
+                prevalence = np.clip((cc_predictions - fpr) / (tpr - fpr), 0, 1)
+                
             prevs.append(prevalence)
         prevalence = np.median(prevs)
         return np.asarray([1 - prevalence, prevalence])

{mlquantify-0.1.11 → mlquantify-0.1.13}/mlquantify/adjust_counting/_base.py

@@ -239,7 +239,10 @@ class BaseAdjustCount(AggregationMixin, BaseQuantifier):
     def aggregate(self, predictions, train_predictions, y_train_values):
         """Aggregate predictions and apply matrix- or rate-based bias correction."""
         self.classes_ = check_classes_attribute(self, np.unique(y_train_values))
-        predictions = validate_predictions(self, train_predictions)
+        
+        predictions = validate_predictions(self, predictions)
+        train_predictions = validate_predictions(self, train_predictions)
+        
         prevalences = self._adjust(predictions, train_predictions, y_train_values)
         prevalences = validate_prevalences(self, prevalences, self.classes_)
         return prevalences

{mlquantify-0.1.11 → mlquantify-0.1.13}/mlquantify/meta/_classes.py

@@ -661,7 +661,7 @@ class QuaDapt(MetaquantifierMixin, BaseQuantifier):
     def __init__(self, 
                  quantifier,
                  measure="topsoe", 
-                 merging_factors=(0.1, 1.0, 0.2)):
+                 merging_factors=np.arange(0.1, 1.0, 0.2)):
         self.quantifier = quantifier
         self.measure = measure
         self.merging_factors = merging_factors
@@ -701,10 +701,7 @@ class QuaDapt(MetaquantifierMixin, BaseQuantifier):
         
         self.classes = self.classes if hasattr(self, 'classes') else np.unique(train_y_values)
 
-        moss = QuaDapt.MoSS(1000, 0.5, m)
-
-        moss_scores = moss[:, :2]
-        moss_labels = moss[:, 2]
+        moss_scores, moss_labels = self.MoSS(1000, 0.5, m)
 
         prevalences = self.quantifier.aggregate(predictions,
                                                 moss_scores,
@@ -721,9 +718,9 @@ class QuaDapt(MetaquantifierMixin, BaseQuantifier):
         distances = []
         
         for mf in MF:
-            scores = QuaDapt.MoSS(1000, 0.5, mf)
-            pos_scores = scores[scores[:, 2] == 1][:, :2]
-            neg_scores = scores[scores[:, 2] == 0][:, :2]
+            scores, labels = self.MoSS(1000, 0.5, mf)
+            pos_scores = scores[labels == 1][:, 1]
+            neg_scores = scores[labels == 0][:, 1]
             
             best_distance = self._get_best_distance(predictions, pos_scores, neg_scores)
             
@@ -772,14 +769,27 @@ class QuaDapt(MetaquantifierMixin, BaseQuantifier):
         .. [1] Maletzke, A., Reis, D. dos, Hassan, W., & Batista, G. (2021).
         Accurately Quantifying under Score Variability. 2021 IEEE International Conference on Data Mining (ICDM), 1228-1233. https://doi.org/10.1109/ICDM51629.2021.00149
         """
-        p_score = np.random.uniform(size=int(n * alpha)) ** m
-        n_score = 1 - (np.random.uniform(size=int(round(n * (1 - alpha), 0))) ** m)
-        scores = np.column_stack(
-            (np.concatenate((p_score, n_score)), 
-             np.concatenate((p_score, n_score)), 
-             np.concatenate((
-                 np.ones(len(p_score)), 
-                 np.full(len(n_score), 0))))
+        if isinstance(alpha, list):
+            alpha = float(alpha[1])
+            
+        n_pos = int(n * alpha)
+        n_neg = n - n_pos
+        
+        # Scores positivos
+        p_score = np.random.uniform(size=n_pos) ** merging_factor
+        # Scores negativos
+        n_score = 1 - (np.random.uniform(size=n_neg) ** merging_factor)
+        
+        # Construção dos arrays de features (duas colunas iguais)
+        moss = np.column_stack(
+            ( 
+                1 - np.concatenate((p_score, n_score)),
+                np.concatenate((p_score, n_score)),
+                np.int16(np.concatenate((np.ones(len(p_score)), np.full(len(n_score), 0))))
+            )
         )
-        return scores
+        
+        scores = moss[:, :2]
+        labels = moss[:, 2].astype(np.int16)
+        return scores, labels
         

{mlquantify-0.1.11 → mlquantify-0.1.13}/mlquantify/mixture/_classes.py

@@ -308,8 +308,10 @@ class SMM(AggregativeMixture):
         mean_pos = np.mean(pos_scores)
         mean_neg = np.mean(neg_scores)
         mean_test = np.mean(predictions)
-        
-        alpha = (mean_test - mean_neg) / (mean_pos - mean_neg)
+        if mean_pos - mean_neg == 0:
+            alpha = mean_test
+        else:
+            alpha = np.clip((mean_test - mean_neg) / (mean_pos - mean_neg), 0, 1)
         return alpha, None
 
 

{mlquantify-0.1.11 → mlquantify-0.1.13}/mlquantify/multiclass.py

@@ -337,7 +337,7 @@ class BinaryQuantifier(MetaquantifierMixin, BaseQuantifier):
         classes = np.unique(args_dict["y_train"])
         qtf.strategy = getattr(qtf, "strategy", "ovr")
 
-        if hasattr(qtf, "binary") and qtf.binary:
+        if (hasattr(qtf, "binary") and qtf.binary) or len(classes) <= 2:
             return qtf._original_aggregate(*args_dict.values())
 
         if qtf.strategy == "ovr":

{mlquantify-0.1.11 → mlquantify-0.1.13/mlquantify.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mlquantify
-Version: 0.1.11
+Version: 0.1.13
 Summary: Quantification Library
 Home-page: https://github.com/luizfernandolj/QuantifyML/tree/master
 Maintainer: Luiz Fernando Luth Junior

mlquantify-0.1.11/VERSION.txt

@@ -1 +0,0 @@
-0.1.11