def run_testsuite(train_X, train_Y):
print("\nSize of training set X: {}".format(train_X.shape))
print("Size of training set Y: {}".format(train_Y.shape))
print("\n+++++Show method 1++++")
train_method1(train_X, train_Y)
print("\n+++++Show method 2++++")
train_method2(train_X, train_Y)
print("\n+++++Show method 3++++")
train_method3(train_X, train_Y)
print("\n+++++Show method 4++++")
train_method4(train_X, train_Y)
print("\n+++++Show method 5++++")
train_method5(train_X, train_Y)
print("\n+++++Show method 6++++")
#train_method6(train_X, train_Y)
# uncomment this if you want to show contour of error graph
#plot_surface_error(train_X, train_Y)
print("\n+++++Show method 7++++")
FX_List = train_method7(train_X, train_Y)
am.visualize(train_X, train_Y, FX_List)
def test_polynomial(X, train_Y ,title, xlabel, ylabel):
# Preprocessing data
scaler_X = preprocessing.StandardScaler().fit(X)
scaler_Y = preprocessing.StandardScaler().fit(train_Y)
X__ = scaler_X.transform(X)
train_Y = scaler_Y.transform(train_Y)
#print(X__[0:5])
#print(train_Y[0:5])
# Generate polynomial features (2 degree).
degree = 2 # You can change to degree 3, 4, 5 ant other at here
poly = PolynomialFeatures(degree)
# output for degree 2 = [1, x, x^2]
# output for degree 3 = [1, x, x^2, x^3]
train_X = poly.fit_transform(X__)
#print(train_X[0:5])
# remove 1 value from array (index 0)
train_X = np.delete(train_X, [0], 1)
#print(train_X[0:5])
assert train_X.shape == (len(train_X), degree) # (xxx, degree)
assert train_Y.shape == (len(train_Y), 1) # (xxx, 1)
print("\n+++++ Example 1++++")
predict = predict_example1(train_X, train_Y)
show_graph(X,
scaler_Y.inverse_transform(train_Y),
scaler_Y.inverse_transform(predict)
,title, xlabel, ylabel)
print("\n+++++ Example 2++++")
predict = predict_example2(train_X, train_Y)
print("\n+++++ Example 3++++")
predict = predict_example3(train_X, train_Y)
print("\n+++++ Example 4++++")
predict = predict_example4(train_X, train_Y)
print("\n+++++ Example 5++++")
predictList, accuracyList, lossList = predict_example5(train_X, train_Y)
am.visualize(X,
scaler_Y.inverse_transform(train_Y),
scaler_Y.inverse_transform(predictList),
accuracyList, lossList, title=title)
def test_one_input(X, train_Y ,title, xlabel, ylabel):
# Preprocessing data
scaler_X = preprocessing.StandardScaler().fit(X)
scaler_Y = preprocessing.StandardScaler().fit(train_Y)
train_X = scaler_X.transform(X)
train_Y = scaler_Y.transform(train_Y)
#print(X__[0:5])
#print(train_Y[0:5])
assert train_X.shape == train_Y.shape
print("\n+++++ Example 1++++")
predict = predict_example1(train_X, train_Y)
show_graph(X,
scaler_Y.inverse_transform(train_Y),
scaler_Y.inverse_transform(predict)
,title, xlabel, ylabel)
print("\n+++++ Example 2++++")
predict = predict_example2(train_X, train_Y)
print("\n+++++ Example 3++++")
predict = predict_example3(train_X, train_Y)
print("\n+++++ Example 4++++")
predict = predict_example4(train_X, train_Y)
print("\n+++++ Example 5++++")
predictList, accuracyList, lossList = predict_example5(train_X, train_Y)
am.visualize(X,
scaler_Y.inverse_transform(train_Y),
scaler_Y.inverse_transform(predictList),
accuracyList, lossList, title=title)
plot_surface_error(train_X, train_Y)
print("\n+++++ Example 6++++")
predict = predict_example6(train_X, train_Y)
print("\n+++++ Example 7++++")
predict = predict_example7(train_X, train_Y)
dfNorm = df.copy() dfNorm[x_column_names] = dfNorm[x_column_names] / X1X2_mean train_X1X2 = dfNorm[x_column_names] # X1, X2 training set train_Y = dfNorm[y_column_name].reshape(-1, 1) # Y (Output) training set C_List = train_method(train_X1X2, train_Y) classA, classB = seperateClass(dfNorm) X1A, X2A = splitFeature(classA) X1B, X2B = splitFeature(classB) # margin = w0 + w1*X1 + w2*X2 # margin/w2 = w0/-w2 + (w1/-w2)*X1 - X2 # ถ้าให้ 0 = w0/w2 + (w1/w2)*X1 - X2 มันคือสมการเส้นตรง ที่ทำนาย X2 # w0/-w2 + (w1/-w2)*X1 -> the decision boundary Equation X1_norm = dfNorm['X1'] X2_norm = dfNorm['X2'] FX_List = getDecisionFunc(X1_norm, C_List) am.visualize(X1A, X2A, X1B, X2B, X1_norm, X2_norm, FX_List) #plot2Class(classA, classB, X1_norm, FX_List[len(FX_List)-1]) # normalize classA and classB #w0, w1, w2 = C_List[len(C_List)-1] #plot2Class(classA, classB, X1, FX[len(FX)-1]) # normalize classA and classB # for visualization #plt.plot(data_X, Y, 'bs', data_X, fx_final, 'r-') #plt.show()