Создание практического задания. Выполнение задач кластеризации разными способами

flats.csv

+id,number_of_rooms,s_squared,price
+1,1,29,5.58
+2,1,34,4.81
+3,1,46,6.44
+4,1,51,7.95
+5,2,59,7.31
+6,2,64,7.88
+7,2,69,8.22
+8,2,51,4.49
+9,2,70,5.3
+10,3,60,8.7
+11,3,84,9.14
+12,3,89,10.18
+13,3,68,5.55
+14,3,76,5.99
+15,3,68,6.2
+16,3,115,9.7
+17,4,112,11.59
+18,4,166,28
+19,4,96,33
+20,3,113,16.5
+21,3,201,37.5
+

kmeans.py

+from pandas import read_csv
+from sklearn.cluster import KMeans
+
+# Оценка разным количеством кластеров
+estimators = [('k_means_iris_2', KMeans(n_clusters=2)),
+('k_means_iris_3', KMeans(n_clusters=3)),
+('k_means_iris_4', KMeans(n_clusters=4))]
+
+data = read_csv('flats.csv', index_col='id', delimiter=',')[:]
+
+x = data.iloc[0:20,[1,2]]
+
+for name, est in estimators:
+    print(name)
+    est.fit(x)
+    print(est)
+    print("Centers")
+    print(est.cluster_centers_)
+    print("Dataset labels")
+    print(est.labels_)
+    # проверяем принадлежность точки 80, 2
+    pr = [[80,2]]
+    pred = est.predict(pr)
+    print("Predicting")
+    print(f"Point {pr[0]} in cluster: {pred[0]}")

knn.py

+import scipy
+import numpy as np
+import pandas as pd
+import random
+import matplotlib.pyplot as plot
+from sklearn.neighbors import NearestNeighbors
+
+size = 100
+points = []
+for i in range(size):
+    points.append([random.randint(0,100), random.randint(0,100)])
+
+pt: pd.DataFrame = pd.DataFrame(points)
+x = pt.iloc[:,0]
+y = pt.iloc[:,1]
+plot.scatter(x,y)
+
+data = np.array(pt.iloc[:,:])
+nbs:NearestNeighbors = NearestNeighbors(n_neighbors=2, algorithm='ball_tree')
+fd = nbs.fit(data)
+
+new_point = [ [ random.randint(0,100), random.randint(0,100) ] ]
+distances, indices = nbs.kneighbors(new_point)
+print(f"Point={new_point}, Dist={distances} IND={indices}")
+
+plot.scatter([new_point[0][0]],[new_point[0][1]],edgecolors=[0,0,1])
+plot.show()

titanicknn.py

+import pandas as pd
+
+from sklearn.model_selection import train_test_split
+from sklearn.metrics import accuracy_score
+
+from sklearn import neighbors
+
+train_data = pd.read_csv('train.csv')
+test_data = pd.read_csv('test.csv')
+
+label=train_data.iloc[0:890,0]
+test_d=train_data.iloc[0:890,[1,3,4]] # 1,3,4 PClass,Sex,Age
+testdat=test_data.iloc[0:267,[1,3,4]] # 1,3,4 PClass,Sex,Age
+
+# Заполняем пустые записи
+test_d = test_d.fillna(0)
+testdat = testdat.fillna(0)
+
+x = [test_d, testdat]
+# Замена названия пола на идентификатор
+for change in x:
+    change['sex'] = change['sex'].map({'female':0, 'male': 1}).astype(int)
+
+# Обучающий набор
+train_data,test_data,train_labels,test_labels=train_test_split(test_d,label,random_state=7,train_size=0.7)
+knn = neighbors.KNeighborsClassifier(weights='distance', n_neighbors=5)
+# Тренировка модели
+knn.fit(test_d, label)
+# Проверка на тестовом наборе
+predictions = knn.predict(testdat)
+print(f"Точность={accuracy_score(test_labels, predictions)}")
+