[Course] Advanced Computer Programming, Homework, week 15, scikit-learn

scikit-learn example

import numpy as np
from sklearn import metrics
from sklearn import datasets
from sklearn.model_selection import cross_validate
from sklearn.naive_bayes import GaussianNB
from sklearn.svm import SVC
from sklearn.ensemble import RandomForestClassifier

data = datasets.make_classification(n_samples=2000, n_features=10)
data[0][:5]

array([[ 1.94224588, -0.90004129, -1.29373233, -2.68480168,  0.33210151,
         0.12402016, -0.85576276,  1.17830754,  0.49223751,  1.96238429],
       [ 0.12985395,  0.54406644, -0.35623636,  0.42980539, -0.36361682,
         0.43473325, -0.66051895, -0.45035764, -1.12449465, -0.82254233],
       [-0.38586157,  0.82526339, -1.11620907, -0.78525796, -0.40823213,
         0.13700943,  0.24143533, -1.02235791,  1.03706833, -1.50767359],
       [-0.48923486,  0.13611755,  0.70740195, -1.74092075, -1.15887133,
         0.96963557, -0.94061012, -2.04747774, -1.61738509, -0.41771673],
       [-0.54016059,  0.13232637,  0.62608625, -0.81078144, -0.31259478,
        -0.10926979,  0.78459312, -1.09810794,  1.88645398, -0.11686013]])

data[1][:5]

array([0, 0, 0, 0, 1])

clfs = [GaussianNB(),
        SVC(C=0.1, kernel='rbf', gamma=0.1),
        RandomForestClassifier(n_estimators=100)]

scoring = ['f1_micro', 'f1_macro']

for clf in clfs:
    scores = cross_validate(clf, data[0], data[1], scoring=scoring, cv=10)
    print('--------------------')
    print(str(clf))
    print()
    print('micro: ')
    print(scores['test_f1_micro'])
    print('macro: ')
    print(scores['test_f1_macro'])
    print('ave: ', np.mean(scores['test_f1_micro']), np.mean(scores['test_f1_macro']))

--------------------
GaussianNB(priors=None)

micro: 
[0.90547264 0.91       0.905      0.96       0.925      0.93
 0.94       0.905      0.945      0.94974874]
macro: 
[0.90546328 0.90985577 0.90499762 0.95998399 0.92498312 0.92997199
 0.93994595 0.90497862 0.94498762 0.949717  ]
ave:  0.9275221380534514 0.9274884968705537
--------------------
SVC(C=0.1, cache_size=200, class_weight=None, coef0=0.0,
  decision_function_shape='ovr', degree=3, gamma=0.1, kernel='rbf',
  max_iter=-1, probability=False, random_state=None, shrinking=True,
  tol=0.001, verbose=False)

micro: 
[0.91044776 0.945      0.93       0.985      0.935      0.965
 0.96       0.945      0.965      0.9798995 ]
macro: 
[0.91026786 0.94498762 0.92993694 0.98499962 0.93498537 0.96497811
 0.959996   0.9449656  0.96499912 0.97989899]
ave:  0.9520347258681466 0.9520015248604876
--------------------
RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',
            max_depth=None, max_features='auto', max_leaf_nodes=None,
            min_impurity_decrease=0.0, min_impurity_split=None,
            min_samples_leaf=1, min_samples_split=2,
            min_weight_fraction_leaf=0.0, n_estimators=100, n_jobs=1,
            oob_score=False, random_state=None, verbose=0,
            warm_start=False)

micro: 
[0.92039801 0.945      0.935      0.96       0.935      0.965
 0.945      0.95       0.965      0.97487437]
macro: 
[0.9202381  0.94499862 0.93492028 0.95998399 0.93499837 0.96499212
 0.94499862 0.94995496 0.96499912 0.97487437]
ave:  0.9495272381809545 0.9494958570594563

采用 micro-f1 和 macro-f1 指标评估分类器性能，采用十折交叉验证。通过上述测试，发现在随机构造的这个二分类问题中，SVM在测试集上的效果最佳。