Add data cleaning scenario

examples/datasets/taxes_2.csv

@@ -9,3 +9,5 @@ Texas,1000,0.15
 Texas,2000,0.25
 Texas,3000,0.3
 Texas,5000,0.05
+Texas,6000,0.04
+Texas,7000,0.03

examples/expert/data_cleaning.py

@@ -0,0 +1,90 @@
+import desbordante as db
+import networkx as nx
+import matplotlib.pyplot as plt
+import time
+
+
+class DataCleaner:
+    def __init__(self, violations):
+        self.graph = {}
+        for v1, v2 in violations:
+            self.graph[v1] = self.graph.get(v1, []) + [v2]
+            self.graph[v2] = self.graph.get(v2, []) + [v1]
+        self.nodes = list(self.graph.keys())
+
+
+    def __remove_most_common(self):
+        max_key = self.nodes[0]
+        max_len = len(self.graph[max_key])
+        for key, val in self.graph.items():
+            if len(val) > max_len:
+                max_key = key
+                max_len = len(val)
+
+        for node in self.graph[max_key]:
+            self.graph[node].remove(max_key)
+        self.graph.pop(max_key)
+        self.nodes.remove(max_key)
+
+    def __is_connected(self):
+        if len(self.graph) == 0:
+            raise ValueError("Graph is empty")
+
+        nodes = list(self.graph.keys())
+        visited = dict.fromkeys(nodes, False)
+        stack = [nodes[0]]
+        while len(stack) != 0:
+            cur = stack.pop()
+            if not visited[cur]:
+                neighbours = self.graph[cur]
+                stack.extend(neighbours)
+            visited[cur] = True
+
+        return all([val for _, val in visited.items()])
+
+    def clean(self, violations):
+        while self.__is_connected():
+            self.__remove_most_common()
+
+    def draw(self):
+        G = nx.Graph()
+        G.add_nodes_from(self.nodes)
+        for node, neighbours in self.graph.items():
+            [G.add_edge(node, neighbour) for neighbour in neighbours]
+        print(self.graph)
+        nx.draw(G, with_labels=True)
+        plt.show(block=True)
+
+
+def verify_dc(verificator, dc):    
+    print("Algo execution start")
+
+    algo_start = time.time()
+    verificator.execute(denial_constraint=dc)
+
+    print("Algo execution end")
+
+    result = verificator.dc_holds()
+    total_time = time.time() - algo_start
+
+    print("DC " + dc + " holds: " + str(result))
+    print("Total time elapsed: " + str(total_time))
+
+
+def main():
+    table = '../datasets/taxes_2.csv'
+    dc = "!(t.0 == s.0 and t.1 > s.1 and t.2 < s.2)"
+
+    verificator = db.dc_verification.algorithms.Default()
+    verificator.load_data(table=(table, ',', True))
+
+    verify_dc(verificator, dc)
+    violations = verificator.get_violations()
+    cleaner = DataCleaner(violations)
+
+    cleaner.draw()
+    cleaner.clean(violations)
+    cleaner.draw()
+
+if __name__ == "__main__":
+    main()