def find_colors(train_Y):
cs = []
for val in train_Y:
if val == 0:
cs.append('c')
else:
cs.append('r')
return cs
# find_params_dbscan(X,Y,1)
X1, Y1 = SMOTE(k_neighbors=40).fit_sample(X, Y)
X2, Y2 = CCR(energy=3).fit_sample(X, Y)
X3, Y3 = dbscan_based.DbscanBasedOversample(eps=1,
min_pts=3,
outline_radio=0.9).fit_sample(
X, Y)
plt.rcParams['figure.figsize'] = (5.0, 5.0)
plt.scatter(X[:, 0], X[:, 1], c=find_colors(Y))
plt.axis('off', )
plt.xticks([])
plt.yticks([])
plt.savefig('myplot20_0.png')
#ax0.set_title('Original dataset')
plt.show()
plt.scatter(X1[:, 0], X1[:, 1], c=find_colors(Y1))
#ax1.set_title('CCR')
plt.axis('off')
plt.xticks([])
plt.yticks([])
'adult': (1.6, 3),
'breast-cancer-wisconsin': (0.5, 3),
'haberman': (0.14, 3)
}
eps_es = np.linspace(0.05, 5, 21)
# print(eps
min_pts = [2, 3, 5, 7]
precisions, recall, f1, auc_score = [], [], [], []
for eps in eps_es:
model = KNeighborsClassifier(n_neighbors=3)
list1 = compare_different_oversample_method(
model,
dbscan_based.DbscanBasedOversample(eps=eps,
min_pts=3,
filter_majority=False), X, Y)
# np.array([round(precision, 3), round(recall, 3), round(f1, 3),round(auc_score,3)])
precisions.append(list1[0])
recall.append(list1[1])
f1.append(list1[2])
auc_score.append(list1[3])
plt.plot(
eps_es,
f1,
linestyle='--', # 折线类型
linewidth=2, # 折线宽度
color='c', # 折线颜色
marker='^', # 点的形状
markersize=10, # 点的大小
#model = DecisionTreeClassifier(max_depth=5, min_samples_split=3)
# from sklearn.naive_bayes import GaussianNB
# model=GaussianNB()
from sklearn.linear_model import LogisticRegression
model = LogisticRegression()
model_name = 'LR'
i = 4
import time
time1=time.time()
print()
res_list1 = compare_different_oversample_method(model, None, X, Y)
time2=time.time()
print('没有过采样计算完成,用时:{}'.format(time2-time1))
res_list2 = compare_different_oversample_method(model, dbscan_based.DbscanBasedOversample(eps=eps, min_pts=min_pts,
outline_radio=0.5,
noise_radio=0.1,
# fit_outline_radio=False,
filter_majority=False), X,
Y)
print('dbscan计算完成')
res_list3 = compare_different_oversample_method(model, CCR(), X, Y)
print('ccr计算完成')
res_list4 = compare_different_oversample_method(model, 'smote', X, Y)
print('smote计算完成')
res_list5 = compare_different_oversample_method(model, 'borderline_smote', X, Y)
print('borderline_smote计算完成')
res_list6 = compare_different_oversample_method(model, 'adasyn', X, Y)
print('adasyn计算完成')
res_list7 = compare_different_oversample_method(model, 'SMOTE_ENN', X, Y)
print('SMOTE_ENN计算完成')
print_result('None', res_list1)
def find_params_dbscan(train_X, train_Y, c):
#0.36 2
classifier = DBSCAN(eps=0.6, min_samples=10)
X = train_X[train_Y == c]
C = classifier.fit_predict(X)
print('簇的个数:{}'.format(max(C) + 1))
print(Counter(C))
plt.scatter(X[:, 0], X[:, 1], c=C)
plt.show()
if __name__ == '__main__':
plt.scatter(X[:, 0], X[:, 1], c=Y)
plt.show()
#find_params_dbscan(X,Y,c=1)
X1, Y1 = dbscan_based.DbscanBasedOversample(eps=0.8,
min_pts=5).fit_sample(X, Y)
X2, Y2 = SMOTE(k_neighbors=10).fit_sample(X, Y)
plt.rcParams['figure.figsize'] = (13.0, 4.0)
fig = plt.figure()
ax1 = fig.add_subplot(1, 3, 1)
ax1.scatter(X[:, 0], X[:, 1], c=Y)
plt.axis('off')
plt.xticks([])
plt.yticks([])
ax2 = fig.add_subplot(1, 3, 2)
ax2.scatter(X1[:, 0], X1[:, 1], c=Y1)
plt.axis('off')
plt.xticks([])
plt.yticks([])
ax3 = fig.add_subplot(1, 3, 3)