def showModel(self):
self.load_file = QFileDialog.getOpenFileName(
self, '파일열기', '', '(*.pkl *.h5);; (*.pkl);; (*.h5)')
if self.load_file[0] == '':
return
try:
if self.load_file[0][-3:] == 'pkl':
self.tabWidget.setCurrentIndex(0)
self.model = ARIMAResults.load(self.load_file[0])
self.name = self.load_file[0][::-1]
self.res_name = ''
for i in self.name:
if i == '/':
break
else:
self.res_name = i + self.res_name
else:
self.model = load_model(load_file[0])
except:
print(self.load_file)
if self.load_file[0][-2:] == 'h5':
self.tabWidget.setCurrentIndex(1)
self.model = tf.keras.models.load_model(self.load_file[0])
self.name = self.load_file[0][::-1]
self.res_name = ''
for i in self.name:
if i == '/':
break
else:
self.res_name = i + self.res_name
# print(self.model)
# Model_name = Model_name + '.h5'
# model = tf.keras.models.load_model(Model_path + '/' + Model_name)
QMessageBox.about(self, "Alert", "성공적으로 불러왔습니다.")
self.model_name.setText(self.res_name)
def predict_sarimax(**kwargs) -> dict:
with open('utils/last_exogs.pkl', 'rb') as f:
exogs = pickle.load(f)
exogs = pd.DataFrame.from_dict(exogs)
exogs = exogs[['InTransit', 'Export', 'Import', 'Transit']]
models = {}
for col in cols:
models[col] = ARIMAResults.load(f'utils/models/sarimax_{col}.pkl')
start = models['Export'].fittedvalues.index.max()
exogs.index = [start]
def iteract(start):
start = start + pd.DateOffset(months=1)
tmp = {}
for col in cols:
tmp[col] = models[col].predict(
start, exog=exogs.loc[:, exogs.columns != col])[0]
return start, tmp
if 'next' in kwargs.keys():
for _ in range(args.next):
start, tmp = iteract(start)
exogs.loc[start] = tmp
return exogs.iloc[1:].to_dict('series')
elif 'start' in kwargs.keys() and 'end' in kwargs.keys():
while str(start)[:7] != kwargs['end']:
start, tmp = iteract(start)
exogs.loc[start] = tmp
return exogs.loc[kwargs['start']:kwargs['end']].to_dict('series')
def load(self, model_path, scaler_path=None):
'''
Loads the model from disk
:param model_path: Path to saved model
'''
if os.path.splitext(model_path)[1] != '.pkl':
raise Exception('Model file path for ' + self.name +
' must have ".pkl" extension.')
self.model = ARIMAResults.load(model_path)
return
def predict():
dataset = read_csv('SmtExpMngr/models/train_data.csv', header=None, index_col=0, parse_dates=True, squeeze=True)
X = dataset.values.astype('float32')
history = [x for x in X]
months_in_year = 1
validation = read_csv('validation.csv', header=None, index_col=0, parse_dates=True, squeeze=True)
y = validation.values.astype('float32')
# load model
model_fit = ARIMAResults.load('/home/termi/beproject/SmtExpMngr/models/model.pkl')
bias = numpy.load('/home/termi/beproject/SmtExpMngr/models/model_bias.npy')
# make first prediction
predictions = list()
yhat = float(model_fit.forecast())
yhat = bias + inverse_difference(history, yhat, months_in_year)
predictions.append(yhat)
history.append(y[0])
return yhat
def get(self,period):
print("Period to predict : ",period)
# Connexion au client HDFS
client = InsecureClient(url='http://namenode:9870', user='******')
# Vérification de la présence du modèle sauvegardé sur HDFS
if client.status(model_hdfs_remote_path + model_name , strict=False) != None:
# load model
client.download(model_hdfs_remote_path+model_name, model_local_path, overwrite=True)
model_fit = ARIMAResults.load(model_local_path + model_name)
# Dataset pour l'évaluation
df = get_data_cassandra()
print(df.head())
X = df['total_estimated_load'].values
start_index = len(X)
end_index = start_index + int(period)
forecast = model_fit.predict(start=start_index, end=end_index)
#df['date_est_load'] = df['date_est_load'].apply(pd.Timestamp)
day = df['date_est_load'].values[-1].date()
print(day)
print(type(day))
day += datetime.timedelta(days=1)
res = {}
for yhat in forecast:
res[day.strftime("%d/%m/%Y")] = yhat
day += datetime.timedelta(days=1)
return res
return "Service has been stopped"
# load the finalized model and make a prediction
from statsmodels.tsa.arima.model import ARIMAResults
from math import exp
from math import log
import numpy
# invert box-cox transform
def boxcox_inverse(value, lam):
if lam == 0:
return exp(value)
return exp(log(lam * value + 1) / lam)
model_fit = ARIMAResults.load('model.pkl')
lam = numpy.load('model_lambda.npy')
yhat = model_fit.forecast()[0]
yhat = boxcox_inverse(yhat, lam)
print('Predicted: %.3f' % yhat)
def predict(self, model: ARIMAResults, start, end):
return model.predict(start, end)
def forecast(self, model: ARIMAResults, n_steps):
return model.forecast(n_steps)
last_exogs[key] = [df.iloc[-1][key]]
with open('utils/last_exogs.pkl', 'wb') as f:
pickle.dump(last_exogs, f)
predictions = {}
cols = ['InTransit', 'Export', 'Import', 'Transit']
if args.predict:
start, end = args.predict.split(':')
print(f'Прогнозирование в диапазоне от {start} до {end}.')
try:
if args.sarimax:
predictions = predict_sarimax(start=start, end=end)
else:
for col in cols:
model = ARIMAResults.load(f'utils/models/sarima_{col}.pkl')
predictions[col] = model.predict(start, end)
except FileNotFoundError:
print(
'Перед прогнозированием нужно создать модель. Запустите программу с флагом -r/--retrain и '
'укажите путь до таблицы с исходными данными в формате CSV или XLSX'
)
exit(1)
if args.next:
print(f'Прогнозирование на ближайшие месяцы: {args.next}.')
try:
if args.sarimax:
predictions = predict_sarimax(next=args.next)
else:
for col in cols: