Create prism.py

Author: Vijayaa21

Python/algorithms/mathematical/prism.py

@@ -0,0 +1,108 @@
+"""
+Data Structure: PRISM Rule-Based Classifier
+
+Description:
+    PRISM is a rule-based machine learning algorithm used for classification. 
+    It generates simple if-then rules for each class by analyzing attribute-value pairs.
+
+Use Case:
+    Useful in applications requiring interpretable classification rules,
+    such as medical diagnosis, decision support systems, or expert systems.
+
+Time Complexity:
+    O(N * M * V), where
+    - N = number of instances
+    - M = number of attributes
+    - V = average number of attribute values
+
+Space Complexity:
+    O(R), where R = number of generated rules.
+"""
+
+from collections import Counter
+
+class PRISM:
+    def __init__(self, data, attributes, class_label):
+        """
+        Initialize the PRISM classifier.
+        
+        Args:
+            data (list of dict): Dataset as list of instances (dict of attribute: value).
+            attributes (list): List of attribute names.
+            class_label (str): Name of the class attribute.
+        """
+        self.data = data
+        self.attributes = attributes
+        self.class_label = class_label
+        self.rules = []
+
+    def train(self):
+        """Generate rules for each class."""
+        classes = set([d[self.class_label] for d in self.data])
+        for cls in classes:
+            examples = [d for d in self.data if d[self.class_label] == cls]
+            while examples:
+                rule = self._generate_rule(examples, cls)
+                self.rules.append(rule)
+                # Remove examples covered by this rule
+                examples = [ex for ex in examples if not self._matches_rule(ex, rule)]
+
+    def _generate_rule(self, examples, cls):
+        """Generate a single rule for a class."""
+        rule = {}
+        remaining_attrs = self.attributes.copy()
+        while True:
+            best_attr, best_value, best_accuracy = None, None, 0
+            for attr in remaining_attrs:
+                values = set([ex[attr] for ex in examples])
+                for val in values:
+                    covered = [ex for ex in examples if ex[attr] == val]
+                    accuracy = sum(1 for ex in covered if ex[self.class_label] == cls) / len(covered)
+                    if accuracy > best_accuracy:
+                        best_attr, best_value, best_accuracy = attr, val, accuracy
+            if best_attr is None:
+                break
+            rule[best_attr] = best_value
+            remaining_attrs.remove(best_attr)
+            # Filter examples covered by this condition
+            examples = [ex for ex in examples if ex[best_attr] == best_value]
+            if all(ex[self.class_label] == cls for ex in examples):
+                break
+        rule['class'] = cls
+        return rule
+
+    def _matches_rule(self, instance, rule):
+        """Check if an instance matches a rule."""
+        for attr, val in rule.items():
+            if attr == 'class':
+                continue
+            if instance[attr] != val:
+                return False
+        return True
+
+    def predict(self, instance):
+        """Predict class label for a single instance."""
+        for rule in self.rules:
+            if self._matches_rule(instance, rule):
+                return rule['class']
+        return None
+
+
+def main():
+    """Test PRISM implementation."""
+    data = [
+        {'Outlook':'Sunny', 'Temp':'Hot', 'Humidity':'High', 'Wind':'Weak', 'PlayTennis':'No'},
+        {'Outlook':'Sunny', 'Temp':'Hot', 'Humidity':'High', 'Wind':'Strong', 'PlayTennis':'No'},
+        {'Outlook':'Overcast', 'Temp':'Hot', 'Humidity':'High', 'Wind':'Weak', 'PlayTennis':'Yes'},
+        {'Outlook':'Rain', 'Temp':'Mild', 'Humidity':'High', 'Wind':'Weak', 'PlayTennis':'Yes'}
+    ]
+    attributes = ['Outlook', 'Temp', 'Humidity', 'Wind']
+    prism = PRISM(data, attributes, 'PlayTennis')
+    prism.train()
+    print("Generated Rules:", prism.rules)
+    test_instance = {'Outlook':'Rain', 'Temp':'Mild', 'Humidity':'High', 'Wind':'Weak'}
+    print("Prediction:", prism.predict(test_instance))
+
+
+if __name__ == "__main__":
+    main()