add descriptive statistics to product distribution op

flamapy/metamodels/bdd_metamodel/operations/bdd_product_distribution.py

@@ -1,8 +1,7 @@
 import math
 from collections import defaultdict
-from typing import cast
-
-from dd.autoref import Function
+from typing import cast, Any
+import statistics
 
 from flamapy.core.models import VariabilityModel
 from flamapy.metamodels.bdd_metamodel.models import BDDModel
@@ -24,6 +23,29 @@ def get_result(self) -> list[int]:
 
     def product_distribution(self) -> list[int]:
         return self.get_result()
+
+    def descriptive_statistics(self) -> dict[str, Any]:
+        if not self._result:
+            return {}
+        else:
+            return descriptive_statistics(self._result)
+
+
+def descriptive_statistics(prod_dist: list[int]) -> dict[str, Any]:
+    values = [i for i in range(len(prod_dist)) for _ in range(prod_dist[i])]
+    min_dist = min(values)
+    max_dist = max(values)
+    median = statistics.median(values)
+    result = {'Mean': statistics.mean(values),
+              'Standard deviation': statistics.stdev(values),
+              'Median': median,
+              'Median absolute deviation': statistics.median([abs(n - median) for n in values]),
+              'Mode': statistics.mode(values),
+              'Min': min_dist,
+              'Max': max_dist,
+              'Range': max_dist - min_dist}
+    return result
+
 
 
 def product_distribution(bdd_model: BDDModel) -> list[int]:
@@ -58,7 +80,7 @@ def product_distribution(bdd_model: BDDModel) -> list[int]:
 
 
 def get_prod_dist(bdd_model: BDDModel,
-                  node: Function, 
+                  node: Any, 
                   dist: dict[int, list[int]], 
                   mark: dict[int, bool], 
                   complemented: bool) -> None:

flamapy/metamodels/bdd_metamodel/operations/interfaces/feature_inclusion_probability.py

@@ -18,7 +18,7 @@ class FeatureInclusionProbability(Operation):
     The other two core metrics are the number of features the SPL manages, 
     and the number of valid product that can be derived.
 
-    Ref.: [Heradio et al. 2019. Supporting the Statistical Analysis of Variability Models. SPLC. 
+    Ref.: [Heradio et al. 2019. Supporting the Statistical Analysis of Variability Models. 
     (https://doi.org/10.1109/ICSE.2019.00091)]
     """
 

flamapy/metamodels/bdd_metamodel/operations/interfaces/product_distribution.py

@@ -1,4 +1,5 @@
 from abc import abstractmethod
+from typing import Any
 
 from flamapy.core.operations import Operation
 
@@ -10,6 +11,11 @@ class ProductDistribution(Operation):
     It accounts for how many products have no features, one features, 
     two features, ..., all features.
 
+    This operation also provides a descriptive statistics summarizing the product distribution of 
+    a variability model. Concretely, it provides:
+    Mean, Standard deviation, Median, Median absolute deviation, Mode, Min, Max, and Range, of the
+    product distribution of the variability model.
+
     This operation in combination with the Feature Inclusion Probability (FIP) operation
     provides a clear picture of the products' homogeneity (i.e., how much does one product 
     differ from the others).
@@ -18,7 +24,7 @@ class ProductDistribution(Operation):
     The other two core metrics are the number of features the SPL manages, 
     and the number of valid product that can be derived.
 
-    Ref.: [Heradio et al. 2019. Supporting the Statistical Analysis of Variability Models. SPLC. 
+    Ref.: [Heradio et al. 2019. Supporting the Statistical Analysis of Variability Models. 
     (https://doi.org/10.1109/ICSE.2019.00091)]
     """
 
@@ -29,3 +35,7 @@ def __init__(self) -> None:
     @abstractmethod
     def product_distribution(self) -> list[int]:
         pass
+
+    @abstractmethod
+    def descriptive_statistics(self) -> dict[str, Any]:
+        pass

tests/test_bdd_metamodel.py

@@ -68,9 +68,13 @@ def analyze_bdd(bdd_model: BDDModel) -> None:
     #     print(f'Config {i}: {config.get_selected_elements()}')
 
     # BDD product distribution
-    dist = BDDProductDistribution().execute(bdd_model).get_result()
+    dist_op = BDDProductDistribution().execute(bdd_model)
+    dist = dist_op.get_result()
+    ds_dist = dist_op.descriptive_statistics()
     print(f'Product Distribution: {dist}')
     print(f'#Products: {sum(dist)}')
+    for ds, v in ds_dist.items():
+        print(f'{ds}: {v}')
 
     # BDD feature inclusion probabilities
     probabilities = BDDFeatureInclusionProbability().execute(bdd_model).get_result()