fix bugs / clean code and use numpy for fastest code

cluster_ga/genetic.py

@@ -1,136 +1,80 @@
 import numpy as np
-import pandas as pd
-from sklearn.metrics import silhouette_score
-from sklearn import datasets
 import random
+from sklearn.metrics import silhouette_score
 
-class genetic(object):
-  def __init__(self, chromosome, points = None):
-    self.chromosome = chromosome
-    self.points = points
-    self.list_class = list(set(datasets.load_iris().target))
+class GeneticClustering:
+    def __init__(self, chromosome: dict, points: np.array, list_class: list) -> None:
+        self.chromosome = chromosome
+        self.points = points
+        self.list_class = list_class
 
-    self.fitness_scores = self._get_fitness()
+    def get_fitness(self):
+        labels = np.array(list(self.chromosome.values()))
+        unique_labels = np.unique(labels)
 
-  def _get_fitness(self):
-
-    labels = np.array(list((self.chromosome.values())))
-    unique_labels = np.unique(labels)
-
-    if len(unique_labels) < 2:
-      return -1
-
-    fitness_scores = silhouette_score(self.points, labels)
-    return fitness_scores
+        if len(unique_labels) < 2:
+            return -1
 
+        return silhouette_score(self.points, labels)
 
-  def mutation(self):
-    probabality = random.random()
-    if probabality >= 0.95:
-      self._change_labels_to_nearest_center()
-      self._change_random_label()
-      self._assign_nearest_cluster_label()
-      return self.chromosome
+    def mutate(self):
+        if random.random() >= 0.95:
+            self._change_labels_to_nearest_center()
+            self._change_random_label()
+            self._assign_nearest_cluster_label()
+        return self.chromosome
 
-    return self.chromosome
+    def generate(self, parent):
+        new_generation = {}
+        for index in range(len(self.chromosome)):
+            prob = random.random()
 
-  def genrate(self, parent):
-    new_generation = dict()
-    for index in range(len(self.chromosome)):
+            if prob < 0.45:
+                new_generation[index] = self.chromosome[index]
+            elif prob < 0.90:
+                new_generation[index] = parent.chromosome[index]
+            else:
+                new_generation[index] = random.choice(self.list_class)
 
-      probabality = random.random()
-
-      if probabality < 0.45:
-        new_generation[index] =  self.chromosome[index]
-
-      elif probabality < 0.90:
-        new_generation[index] = parent.chromosome[index]
-
-      else:
-        new_generation[index] =  random.choice(self.list_class)
-
-    return genetic(new_generation, self.points)
-
-
-  def _change_labels_to_nearest_center(self):
-    sample_cluster_labels = random.choice(self.list_class)
-    labels_neares_cluster = self._find_labels_neares_cluster(sample_cluster_labels)
-
-    number_of_sample_cluster_labels = list(self.chromosome.values()).count(sample_cluster_labels)
-
-    counter = 1
-    for index in range(len(self.chromosome)):
-      if self.chromosome[index] == sample_cluster_labels:
-        self.chromosome[index] = labels_neares_cluster
-        counter += 1
-      if counter == number_of_sample_cluster_labels // 3:
-        break
-
-  def _find_labels_neares_cluster(self, sample_cluster_labels):
-    sample_cluster = {}
-    _sample_cluster = []
-    for index in range(len(self.chromosome)):
-      if self.chromosome[index] == sample_cluster_labels:
-        _sample_cluster.append(self.points[index])
-    sample_cluster[sample_cluster_labels] = _sample_cluster
-
-    clusters = {}
-    for _label in self.list_class:
-      cluster = []
-      if sample_cluster_labels != _label:
-        for index, label in self.chromosome.items():
-          if label == _label:
-            cluster.append(self.points[index])
-        clusters[_label] = cluster
-
-    center_sample = self._find_cneter_cluster(sample_cluster[sample_cluster_labels])
-
-    centers = []
-    for cluster in clusters:
-      temp = self._find_cneter_cluster(clusters[cluster])
-      centers.append(temp)
+        return GeneticClustering(new_generation, self.points, self.list_class)
 
-    distance = []
-    for center in centers:
-      _distance = self._calcute_distance(center, center_sample)
-      distance.append(_distance)
+    def _change_labels_to_nearest_center(self):
+        sample_cluster_label = random.choice(self.list_class)
+        nearest_cluster_label = self._find_nearest_cluster(sample_cluster_label)
 
-    distance = np.argsort(distance)
-    index_nearest_center = distance[0]
-
-    labels_neareset_cluster = list(clusters)[index_nearest_center]
-    return labels_neareset_cluster
+        sample_cluster_indices = [index for index, label in self.chromosome.items() if label == sample_cluster_label]
+        num_changes = len(sample_cluster_indices) // 3
 
-  def _find_cneter_cluster(self, cluster):
-    data = list(map(np.array, cluster))
-    data = np.array(data)
-    center = np.mean(data, axis=0)
-    return center
+        for idx in sample_cluster_indices[:num_changes]:
+            self.chromosome[idx] = nearest_cluster_label
 
-  def _assign_nearest_cluster_label(self):
-    sample_index = random.choice(list(self.chromosome.keys()))
-
-    _distance = []
-    for index in range(len(self.chromosome)):
-      _distance.append(self._calcute_distance(self.points[index], self.points[sample_index]))
-    _distance = np.argsort(_distance)
-    index_nearest_point = _distance[1]
-    self.chromosome[sample_index] = self.chromosome[index_nearest_point]
+    def _find_nearest_cluster(self, sample_cluster_label):
+        sample_points = self.points[list(self.chromosome.values()) == sample_cluster_label]
+        sample_center = np.mean(sample_points, axis=0)
 
+        # Get centers of all other clusters
+        other_clusters = {label: [] for label in self.list_class if label != sample_cluster_label}
+        for index, label in self.chromosome.items():
+            if label != sample_cluster_label:
+                other_clusters[label].append(self.points[index])
 
-  def _calcute_distance(self, point1, point2):
-    point1 = np.array(point1)
-    point2 = np.array(point2)
-    distance = np.sqrt(np.sum((point1 - point2) ** 2))
-    return distance
+        cluster_centers = {label: np.mean(np.array(points), axis=0) for label, points in other_clusters.items()}
+        
+        distances = {label: np.linalg.norm(sample_center - center) for label, center in cluster_centers.items()}
+        nearest_label = min(distances, key=distances.get)
+        return nearest_label
 
+    def _assign_nearest_cluster_label(self):
+        sample_index = random.choice(list(self.chromosome.keys()))
+        distances = np.linalg.norm(self.points - self.points[sample_index], axis=1)
+        nearest_index = np.argsort(distances)[1]  # 1st closest point
+        self.chromosome[sample_index] = self.chromosome[nearest_index]
 
-  def _change_random_label(self):
+    def _change_random_label(self):
+        sample_index = random.choice(list(self.chromosome.keys()))
+        new_label = random.choice(self.list_class)
 
-    sample_index = random.choice(list(self.chromosome.keys()))
-    sample_labels = random.choice(self.list_class)
+        while new_label == self.chromosome[sample_index]:
+            new_label = random.choice(self.list_class)
 
-    while sample_labels == self.chromosome[sample_index]:
-      sample_labels = random.choice(self.list_class)
-    self.chromosome[sample_index] = sample_labels  
-
+        self.chromosome[sample_index] = new_label