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 getCorrelationDic():
corDic={}
for feature in Atributes:
x,y=createCustomeDataSet([feature])
corDic[feature]=Global.getCorrelation(x, y)
return corDic
