def draw_svm_db(kernel="linear", gamma=1000, C=1.0):
#Train
X_train, X_test, y_train, y_test = mytools.make_fake_data()
clf = SVC(kernel=kernel, gamma=gamma, C=C)
start = time.time()
clf.fit(X_train, y_train)
train_time = time.time() - start
#Predict
start = time.time()
preds = clf.predict(X_test)
test_time = time.time() - start
#Score
accuracy = accuracy_score(y_test, preds)
print ""
print "Kernel: ", kernel
print "C: ", C, " , gamma: ", gamma
print "Accuracy: ", accuracy
print "Training time: ", round(train_time,3)
print "Test time: ", round(test_time,3)
#Visualization
mytools.prettyPicture(clf, X_test, y_test)
plt.show()
y = [round(bumpiness[i]*grade[i]+0.3+0.1) for i in xrange(0,n_points)]
for i in xrange(0,n_points):
if grade[i] > 0.8 and bumpiness[i]>0.8:
y[i] = 1.0
#print y
#split to test and train
X = [[gg, bb] for gg,bb in zip(grade, bumpiness)]
split = int(0.75*n_points)
X_train = X[0:split]
X_test = X[split:]
y_train = y[0:split]
y_test = y[split:]
return X_train, X_test, y_train, y_test
X_train, X_test, y_train, y_test = mytools.make_fake_data()
#scatter plot
xx=[]
yy=[]
colors=[]
for idx, point in enumerate(X):
xx.append(point[0])
yy.append(point[1])
if y[idx] == 1:
colors.append('blue') #fase
else:
colors.append("yellow")
plt.figure(0)
plt.title("Training data")
plt.xlabel('grade')
plt.ylabel('bumpiness')