def analysis():
X, y = get_xy()
X_train, X_test, Y_train, Y_test, = \
train_test_split(X.toarray(), y,
test_size=0.3, random_state=42)
clf = create_learner()
clf.fit(X_train, Y_train)
prob = clf.predict_proba(X_test)[:, 1]
precision, recall, threld = precision_recall_curve(Y_test, prob)
area = auc(recall, precision)
plot_precision_recall(precision, recall, area, clf.__class__.__name__)
## roc
fpr, tpr, roc_thresholds = roc_curve(Y_test, prob)
area = auc(fpr, tpr)
plot_roc(area, tpr, fpr, clf.__class__.__name__)
def analysis():
X, y = get_xy()
X_train, X_test, Y_train, Y_test, = \
train_test_split(X.toarray(), y,
test_size=0.3, random_state=42)
clf = create_learner()
clf.fit(X_train, Y_train)
prob = clf.predict_proba(X_test)[:, 1]
precision, recall, threld = precision_recall_curve(Y_test, prob)
area = auc(recall, precision)
plot_precision_recall(precision, recall, area, clf.__class__.__name__)
## roc
fpr, tpr, roc_thresholds = roc_curve(Y_test, prob)
area = auc(fpr, tpr)
plot_roc(area, tpr, fpr, clf.__class__.__name__)
def analysis_1vsN():
print('============== One Vs Rest ==============')
X_train, Y_train, X_test, Y_test, targetname = create_xy()
clf = create_learner()
clf.fit(X_train, Y_train)
cm = confusion_matrix(Y_test, clf.predict(X_test))
for i in range(3):
print convert2d(cm, i)
yy = np.asarray(Y_test==i, dtype=int)
prob = clf.predict_proba(X_test)[:,i]
## precision-recall
precision, recall, threld = precision_recall_curve(yy, prob)
area = auc(recall, precision)
plot_precision_recall(precision, recall, area, targetname[i])
## roc
fpr, tpr, roc_thresholds = roc_curve(yy, prob)
area = auc(fpr, tpr)
plot_roc(area, tpr, fpr, targetname[i])
