| 1 | #!/usr/bin/env python3 |
| 2 | |
| 3 | # -*- coding: utf-8 -*- |
| 4 | import random |
| 5 | from sklearn import neighbors |
| 6 | from sklearn.cross_validation import train_test_split |
| 7 | from sklearn.datasets import load_iris |
| 8 | irisData = load_iris() |
| 9 | |
| 10 | X = irisData.data |
| 11 | Y = irisData.target |
| 12 | |
| 13 | # print(help(train_test_split)) |
| 14 | X_train, X_test, Y_train, Y_test = train_test_split( |
| 15 | X, Y, test_size=0.3, random_state=random.seed()) |
| 16 | # print(len(X_train)) |
| 17 | # print(len(X_test)) |
| 18 | # print(len(X_train[Y_train == 0])) |
| 19 | # print(len(X_train[Y_train == 1])) |
| 20 | # print(len(X_train[Y_train == 2])) |
| 21 | |
| 22 | nb_voisins = 15 |
| 23 | clf = neighbors.KNeighborsClassifier(nb_voisins) |
| 24 | clf.fit(X_train, Y_train) |
| 25 | # print("kNN prediction on [5.4, 3.2, 1.6, 0.4]:") |
| 26 | # print(clf.predict([[5.4, 3.2, 1.6, 0.4]])) |
| 27 | # print("kNN probability prediction on [5.4, 3.2, 1.6, 0.4]:") |
| 28 | # print(clf.predict_proba([[5.4, 3.2, 1.6, 0.4]])) |
| 29 | print("kNN score on Iris test data:") |
| 30 | print(clf.score(X_test, Y_test)) |
| 31 | print("kNN prediction error(s) on Iris test data:") |
| 32 | Z = clf.predict(X_test) |
| 33 | print(X_test[Z != Y_test]) |