def linear_model_normal(valSet,is_visual):
if valSet==None:
data, label = creatingMat.createDataSet()
X_train=data[0:-100]
Y_train=label[0:-100]
X_test=data[-100:]
Y_test=label[-100:]
else:
X_train = valSet[0]
Y_train = valSet[1]
X_test = valSet[2]
Y_test = valSet[3]
linearModel_normal=LinearModel.myLM()
linearModel_normal.myLMtrain(X_train , Y_train )
Y_result=linearModel_normal.myLMpredict(X_test)
if is_visual==True:
#visualization
makeVisual(Y_test,Y_result,'Linear Regression Using Normal equation')
Global.getAccuracy(Y_test, Y_result, 20)
else:
#----------------------- return Global.getAccuracy(Y_test, Y_result, 20)
return Global.getCorrelation(Y_test, Y_result)
def svr_models(valSet,is_visual,C_cons=1e3):
min_max_scaler = preprocessing.MinMaxScaler()
svr_lin = SVR(kernel='linear', C=C_cons)
if valSet == None:
data, label = creatingMat.createDataSet()
X_train = data[0:-100]
Y_train = label[0:-100]
X_test = data[-100:]
Y_test = label[-100:]
else:
X_train = valSet[0]
Y_train = valSet[1]
X_test = valSet[2]
Y_test = valSet[3]
X_minmax = min_max_scaler.fit_transform(X_train + X_test)
Y_minmax = min_max_scaler.fit_transform(Y_train + Y_test)
X_train_minmax = array(X_minmax[0:-100])
Y_train_minmax = array(Y_minmax[0:-100])
X_test_minmax = array(X_minmax[-100:])
Y_test_minmax = array(Y_minmax[-100:])
Y_train_minmax_1d=[i[0] for i in Y_train_minmax ]
Y_result=svr_lin.fit(X_train_minmax,Y_train_minmax_1d).predict(array(X_test_minmax))
Y_result_new=[[i] for i in Y_result]
F_result = Global.getAccuracy(Y_test_minmax, Y_result_new, 20)
return Global.getCorrelation(Y_test_minmax, Y_result_new)
def makeVisual(Y_test,Y_result,label_string):
Xaxi = range(1, 50)
Yaxi = [Global.getAccuracy(Y_test, Y_result, i) for i in Xaxi ]
plt.axis([0, 70, 0, 52])
plt.xlabel('Success percentage')
plt.ylabel('Error Tolerance')
plt.title('Success percentage Against Error Tolerance ')
plt.plot(Yaxi, Xaxi, color='red', label=label_string, linewidth=2.5, linestyle="-")
plt.legend()
plt.show()
def linear_model_gradient(valSet,is_visual):
if valSet==None:
data, label = creatingMat.createDataSet()
X_train=data[0:-100]
Y_train=label[0:-100]
X_test=data[-100:]
Y_test=label[-100:]
else:
X_train = valSet[0]
Y_train = valSet[1]
X_test = valSet[2]
Y_test = valSet[3]
regr = linear_model.LinearRegression()
Y_result=regr.fit(X_train,Y_train).predict(X_test)
if is_visual==True:
#visualization
makeVisual(Y_test,Y_result,'Linear Regression Using Gradient Descent')
return Global.getAccuracy(Y_test, Y_result, 20)
else:
return Global.getAccuracy(Y_test, Y_result, 20)
def getSuccessValue(featureList, tollerance):
"""
it is for get success ratio of subset
Args: list of subset , error tolerance
Returns: Success rate of corresponding subset
"""
data, label = createCustomeDataSet(featureList)
X_train = data[0:-100]
Y_train = label[0:-100]
X_test = data[100:]
Y_test = label[100:]
regr = linear_model.LinearRegression()
Y_result = regr.fit(X_train, Y_train).predict(X_test)
return Global.getAccuracy(Y_test, Y_result, tollerance)
def waste():
colors = ['red', 'blue', 'green']
labels = {'red':'SVM-Linear', 'blue':'SVM-Polynomial', 'green':'SVM-RBF'}
Xaxi = range(1, 50)
#visualization
plt.axis([0, 70, 0, 52])
plt.xlabel('Success percentage')
plt.ylabel('Error Tolerance')
plt.title('Success percentage Against Error Tolerance ')
results=0
Y_test_minmax=0
for result, colr in zip(results, colors):
Y_test = [[i] for i in result]
Yaxi = [Global.getAccuracy(Y_test_minmax, Y_test, i) for i in Xaxi ]
plt.plot(Yaxi, Xaxi, color=colr, label=labels[colr], linewidth=2.5, linestyle="-")
plt.legend()
plt.show()
