Merge pull request #28 from zStupan/main

Major Refactor + Added metrics

README.md

@@ -57,9 +57,10 @@ data = Dataset("datasets/Abalone.csv")
 # Create a problem:::
 # dimension represents the dimension of the problem;
 # features represent the list of features, while transactions depicts the list of transactions
-# the following 4 elements represent weights (support, confidence, coverage, shrinkage)
-# None defines that criteria are omitted and are, therefore, excluded from the fitness function
-problem = NiaARM(data.dimension, data.features, data.transactions, alpha=1.0, beta=1.0)
+# metrics is a sequence of metrics to be taken into account when computing the fitness;
+# you can also pass in a dict of the shape {'metric_name': <weight of metric in range [0, 1]>};
+# when passing a sequence, the weights default to 1.
+problem = NiaARM(data.dimension, data.features, data.transactions, metrics=('support', 'confidence'), logging=True)
 
 # build niapy task
 task = Task(problem=problem, max_iters=30, optimization_type=OptimizationType.MAXIMIZATION)

examples/basic_run.py

@@ -13,7 +13,7 @@
     # features represent the list of features, while transactions depicts the list of transactions
     # the following 4 elements represent weights (support, confidence, coverage, shrinkage)
     # A weight of 0.0 means that criteria are omitted and are, therefore, excluded from the fitness function
-    problem = NiaARM(data.dimension, data.features, data.transactions, alpha=1.0, beta=1.0, logging=True)
+    problem = NiaARM(data.dimension, data.features, data.transactions, metrics=('support', 'confidence'), logging=True)
 
     # build niapy task
     task = Task(problem=problem, max_iters=30, optimization_type=OptimizationType.MAXIMIZATION)

examples/stats.py

@@ -14,7 +14,7 @@
 if __name__ == '__main__':
     # Load the dataset and run the algorithm
     data = Dataset("datasets/Abalone.csv")
-    problem = NiaARM(data.dimension, data.features, data.transactions, alpha=1.0, beta=1.0)
+    problem = NiaARM(data.dimension, data.features, data.transactions, metrics=('support', 'confidence'))
     task = Task(problem=problem, max_iters=30, optimization_type=OptimizationType.MAXIMIZATION)
     algo = DifferentialEvolution(population_size=50, differential_weight=0.5, crossover_probability=0.9)
     algo.run(task=task)

niaarm/dataset.py

@@ -14,28 +14,26 @@ class Dataset:
          If the file already contains a header row, pass ``header=0`` to override the feature names.
 
     Attributes:
-        data (pd.DataFrame): Data as a pandas Dataframe.
-        transactions (np.ndarray): Transactional data.
+        transactions (pandas.DataFrame): Transactional data.
         header (list[str]): Feature names.
         features (list[Feature]): List of features.
         dimension (int): Dimension of the optimization problem for the dataset.
 
     """
 
     def __init__(self, path, delimiter=',', header=0, names=None):
-        self.data = pd.read_csv(path, delimiter=delimiter, header=header, names=names)
+        self.transactions = pd.read_csv(path, delimiter=delimiter, header=header, names=names)
         if names is None and header is None:
-            self.data.columns = pd.Index([f'Feature{i}' for i in range(len(self.data.columns))])
-        self.header = self.data.columns.tolist()
-        self.transactions = self.data.values
+            self.transactions.columns = pd.Index([f'Feature{i}' for i in range(len(self.transactions.columns))])
+        self.header = self.transactions.columns.tolist()
         self.features = []
         self.__analyse_types()
         self.dimension = self.__problem_dimension()
 
     def __analyse_types(self):
         r"""Extract data types for the data in a dataset."""
         for head in self.header:
-            col = self.data[head]
+            col = self.transactions[head]
 
             if col.dtype == "float":
                 dtype = "float"
@@ -48,8 +46,7 @@ def __analyse_types(self):
                 max_value = col.max()
                 unique_categories = None
             elif col.dtype == 'bool':
-                self.data[head] = self.data[head].astype(int)
-                self.transactions = self.data.values
+                self.transactions[head] = self.transactions[head].astype(int)
                 dtype = 'int'
                 min_value = 0
                 max_value = 1

niaarm/feature.py

@@ -2,7 +2,7 @@
 from typing import Optional
 
 
-@dataclass
+@dataclass(repr=False)
 class Feature:
     r"""Class representing a feature.
 
@@ -11,12 +11,23 @@ class Feature:
        dtype (str): Datatype of the feature.
        min_val (Optional[float]): Minimum value of the feature in the transaction database.
        max_val (Optional[float]): Maximum value of the feature in the transaction database.
-       categories (Optional[list[float]]): Possible categorical feature's values.
+       categories (Optional[list[str]]): Possible categorical feature's values.
 
     """
 
     name: str
     dtype: str
     min_val: Optional[float] = None
     max_val: Optional[float] = None
-    categories: Optional[list[float]] = None
+    categories: Optional[list[str]] = None
+
+    def __repr__(self):
+        string = f'{self.name}('
+        if self.dtype == 'cat':
+            string += f'{self.categories if len(self.categories) != 1 else self.categories[0]})'
+        else:
+            if self.min_val == self.max_val:
+                string += f'{self.min_val})'
+            else:
+                string += f'[{self.min_val}, {self.max_val}])'
+        return string