def run(x):
data_txt = "point_li.txt"
LI.li(data_txt)
data_txt = "point_ls.txt"
LS.ls(data_txt)
data_txt = 'point_ni.txt'
NI.ni(data_txt)
def LSTM_Test(self):
self.canvas.hide()
self.fig3.clear()
stock_file = self.data.loc[:, [self.Date_combobox.currentText()]]
for combo in self.LSTMcomboboxes:
if combo == self.Date_combobox:
continue
stock_file = pd.merge(stock_file,
self.data.loc[:, [combo.currentText()]],
left_index=True,
right_index=True,
how='left')
# window_size = int(self.Window_size_lineText.text())
dense = int(self.predict_periods.text())
graph, result = LS.Test(stock_file, 0.95, dense, dense, self.model)
self.fore = result
self.idx = range(len(self.fore))
try:
self.result_gridLayout.removeWidget(self.canvas3)
except:
pass
self.canvas3 = FigureCanvas(graph)
self.result_gridLayout.addWidget(self.canvas3)
self.canvas3.show()
self.minus_pushButton.hide()
self.plus_pushButton.hide()
def algrun():
global t, list
time_start = time.time()
map = Map.Map()
x, y = Read.Read()
map.path[0] = x
map.path[1] = y
# state == 0: SA, state == 1: LS
state = 0
if state == 0:
# SA(map, T, Coolrate)
alg = SA.SA(map, 100, 0.001)
alg.run()
elif state == 1:
# LS(map, state), state = 0: LS1, state = 1: LS2, state = 2: 2-OPT
alg = LS.LS(map, 2)
alg.run()
list.append(alg.path.distance())
time_end = time.time()
t += (time_end - time_start)
print('totally cost', time_end - time_start, 's')
def lsBtnCallback(self):
global length
global matrix
global isLS_CLICK
global ls_res
v0 = int(self.input_1.get())
v1 = int(self.input_2.get())
if hasattr(self, 'comboBox'):
self.comboBox.destroy()
self.comboBox_label.destroy()
if hasattr(self, 'treeview'):
self.treeview.destroy()
if hasattr(self, 'error'):
self.error.destroy()
if not isLS_CLICK:
isLS_CLICK = True
ls_res = ls.dijkstra(length, copy.deepcopy(matrix), v0)
distance = ls_res[0]
path = ls_res[1]
print(distance)
print(path)
ways = ls.get_ways(v0, v1, path)
print(ways)
if len(ways) == 2:
print('ssss')
if type(ways) == 'NoneType':
self.createError()
return
self.createWays(ways)
drawNetwork.drawResult(ways)
self.createPhoto()
def find_best_model(df, tgt):
lr, ls, dt, dnn = [], [], [], []
for i in range(100):
seed = random.randrange(100)
lr.append(LR.linreg(df, tgt, seed))
ls.append(LS.lasso(df, tgt, seed))
dt.append(DT.dectree(df, tgt, seed))
dnn.append(DNN.nn(df, tgt, seed))
print(pd.DataFrame({'lr': lr}).describe())
print(pd.DataFrame({'ls': ls}).describe())
print(pd.DataFrame({'dt': dt}).describe())
print(pd.DataFrame({'dnn': dnn}).describe())
def LSTM_Train(self):
if self.studying_rate.text() == '':
QMessageBox.about(self, "Alert", "학습 비율을 입력해주세요")
return
elif int(self.studying_rate.text()) < 0:
QMessageBox.about(self, "Alert", "학습 비율을 양수로 입력해주세요")
return
elif self.verification_rate.text() == '':
QMessageBox.about(self, "Alert", "검증 비율을 입력해주세요")
return
elif int(self.verification_rate.text()) < 0:
QMessageBox.about(self, "Alert", "검증 비율을 양수로 입력해주세요")
return
elif self.test_rate.text() == '':
QMessageBox.about(self, "Alert", "테스트 비율을 입력해주세요")
return
elif int(self.test_rate.text()) < 0:
QMessageBox.about(self, "Alert", "테스트 비율을 양수로 입력해주세요")
return
elif self.ModelSize_lineText.text() == '':
QMessageBox.about(self, "Alert", "Model Size 값을 입력해주세요")
return
elif self.WindowSize_lineText.text() == '':
QMessageBox.about(self, "Alert", "Window Size 값을 입력해주세요")
return
elif self.PredictSize_lineText.text() == '':
QMessageBox.about(self, "Alert", "Dense 값을 입력해주세요")
return
elif self.Epochs_linetext.text() == '':
QMessageBox.about(self, "Alert", "Epochs 값을 입력해주세요")
return
elif self.BatchSize_lineText.text() == '':
QMessageBox.about(self, "Alert", "Batch Size 값을 입력해주세요")
return
elif self.Patience_lineText.text() == '':
QMessageBox.about(self, "Alert", "Early Stop 값을 입력해주세요")
return
stock_file = self.data.loc[:, [self.Date_combobox.currentText()]]
for combo in self.LSTMcomboboxes:
if combo == self.Date_combobox:
continue
stock_file = pd.merge(stock_file,
self.data.loc[:, [combo.currentText()]],
left_index=True,
right_index=True,
how='left')
epochs = int(self.Epochs_linetext.text())
batch_size = int(self.BatchSize_lineText.text())
train_ratio = int(self.studying_rate.text()) / (
int(self.studying_rate.text()) +
int(self.verification_rate.text()) + int(self.test_rate.text()))
valid_ratio = int(self.verification_rate.text()) / (
int(self.studying_rate.text()) +
int(self.verification_rate.text()) + int(self.test_rate.text()))
window_size = int(self.WindowSize_lineText.text())
dense = int(self.PredictSize_lineText.text())
Model_size = int(self.ModelSize_lineText.text())
patience = int(self.Patience_lineText.text())
LSTM_lossgraph, LSTM_validgraph_origin, LSTM_validgraph_plus, train_part, rmse, self.LSTM_model = LS.Train(
stock_file, train_ratio, valid_ratio, window_size, dense,
Model_size, patience, epochs, batch_size)
# Loss
self.canvas = FigureCanvas(LSTM_lossgraph)
# 학습 origin
self.canvas2 = FigureCanvas(LSTM_validgraph_origin)
# 학습 확대
self.canvas3 = FigureCanvas(LSTM_validgraph_plus)
self.canvas3.hide()
self.LSTM_Loss_gridLayout.addWidget(self.canvas)
self.LSTM_Valid_gridLayout.addWidget(self.canvas3)
self.LSTM_Valid_gridLayout.addWidget(self.canvas2)
self.Rmse_label.setText(f'RMSE : {rmse}')
self.Train_label.setText(f'학습량 : {train_part}')
self.model_status = 'LSTM'
# we define first the continuous time mass-damper system and # then discretize it using transitions from Hartmut's course from LS import * from scipy import linalg from scipy import integrate # [0 1] # A = [-1 d] corresponds to periodic motion altered by damping d # first we shall observe the velocity of the object and use # output feedback control md_velocity_feedback = LS([[0, 1],[-1, 1]], [[0],[1]], [0, 1], [0]) D1 = md_velocity_feedback.discretize(0.01) # now we observe the position of the object in what would seem a much # trickier control problem md_position_feedback = LS([[0, 1],[-1, 1]], [[0],[1]], [1, 0], [0]) D2 = md_position_feedback.discretize(0.01)
lr.append(LR.linreg(df, tgt, seed))
ls.append(LS.lasso(df, tgt, seed))
dt.append(DT.dectree(df, tgt, seed))
dnn.append(DNN.nn(df, tgt, seed))
print(pd.DataFrame({'lr': lr}).describe())
print(pd.DataFrame({'ls': ls}).describe())
print(pd.DataFrame({'dt': dt}).describe())
print(pd.DataFrame({'dnn': dnn}).describe())
tgt = 'medv'
df = prepro.Data(tgt)
y = df[tgt]
seed = 101
LR.linreg(df, tgt, seed)
LS.lasso(df, tgt, seed)
DT.dectree(df, tgt, seed)
DNN.nn(df, tgt, seed)
#find_best_model(df, tgt)
#PCA.analysis(df)
