def main(argv):
parser = argparse.ArgumentParser()
parser.add_argument("--model", default="lstm", help="Model to train.")
parser.add_argument("--lag", default="12", help="specify number of lags")
parser.add_argument("--file_name",
default="970_1_data.csv",
help="Csv file name")
args = parser.parse_args()
lag = int(args.lag)
config = {"batch": 256, "epochs": 600} # training config
file = args.file_name
model_name = os.path.splitext(file)[0]
X_train, y_train, _, _, _ = process_data(file, lag)
if args.model == 'lstm':
m = model.get_lstm(
[[3, lag], 64, 64,
1]) # changing the input to 3, lags for rain data model
train_model(m, X_train, y_train, "{}_lstm".format(model_name), config)
if args.model == 'gru':
m = model.get_gru([[3, lag], 64, 64, 1])
train_model(
m, X_train, y_train, "{}_gru".format(model_name),
config) # changing the input to 3, lags for rain data model
def main(argv):
parser = argparse.ArgumentParser()
parser.add_argument("--model", default="lstm",
help="Model to train.") #model name
parser.add_argument("--lag", default="12", help="lags") # number of lags
parser.add_argument("--file_name",
default="970_1_data.csv",
help="Csv file name") # file name of the csv
args = parser.parse_args()
lag = int(args.lag)
config = {"batch": 256, "epochs": 600}
file = args.file_name
X_train, y_train, _, _, _ = process_data(file, lag)
if args.model == 'lstm': # training lstm model
X_train = np.reshape(X_train, (X_train.shape[0], X_train.shape[1], 1))
m = model.get_lstm([lag, 64, 64, 1])
train_model(m, X_train, y_train, args.model, config)
if args.model == 'gru': # training gru model
X_train = np.reshape(X_train, (X_train.shape[0], X_train.shape[1], 1))
m = model.get_gru([lag, 64, 64, 1])
train_model(m, X_train, y_train, args.model, config)
if args.model == 'saes': # training saes model
X_train = np.reshape(X_train, (X_train.shape[0], X_train.shape[1]))
m = model.get_saes([lag, 400, 400, 400, 1])
train_seas(m, X_train, y_train, args.model, config)
def main():
parser = argparse.ArgumentParser()
parser.add_argument("--lag", default="12", help="lags")
parser.add_argument("--model", default="970_1_data.csv", help="csv file name for the model")
args = parser.parse_args()
lag = int(args.lag)
model_name = args.model
model_name = os.path.splitext(model_name)[0] # getting model name
# load all the models
lstm = load_model('model/lstm.h5')
gru = load_model('model/gru.h5')
saes = load_model('model/saes.h5')
knn = pickle.load(open('model/knn.model', 'rb'))
xgboost = xgb.Booster({'nthread': 4}) # init model
xgboost.load_model('model/xgb.model')
models = [lstm, gru, saes,knn,xgboost]
names = ['LSTM', 'GRU', 'SAEs','KNN','Xgboost']
lag = 12
file = '970_1_data.csv'
_, _, X_test, y_test, scaler = process_data(file, lag)
print(y_test.shape)
y_test = scaler.inverse_transform(y_test.reshape(-1, 1)).reshape(1, -1)[0]
print(y_test.shape)
y_preds = []
for name, model in zip(names, models):
_, _, X_test, _,_ = process_data(file, lag)
if name == 'SAEs':
X_test = np.reshape(X_test, (X_test.shape[0], X_test.shape[1]))
elif name == 'GRU' or name == 'LSTM':
X_test = np.reshape(X_test, (X_test.shape[0], X_test.shape[1], 1))
elif name == 'Xgboost':
X_test = xgb.DMatrix(X_test)
# file = 'images/' + name + '.png'
# plot_model(model, to_file=file, show_shapes=True)
predicted = model.predict(X_test)
predicted = scaler.inverse_transform(predicted.reshape(-1, 1)).reshape(1, -1)[0]
y_preds.append(predicted[:384]) # append all the prediction results
print(name)
eva_regress(y_test, predicted)
def main():
parser = argparse.ArgumentParser()
parser.add_argument("--lag", default="12", help="lags")
parser.add_argument("--model", default="970_1_data.csv", help="lags")
args = parser.parse_args()
lag = int(args.lag)
model_name = args.model
model_name = os.path.splitext(model_name)[0]
lstm = load_model('model/{}_lstm.h5'.format(model_name))
# gru = load_model('model/{}_gru.h5'.format(args.model))
# saes = load_model('model/{}_saes.h5'.format(args.model))
models = [lstm]
names = ['LSTM']
lag = 12
file = '{}.csv'.format(model_name)
_, _, X_test, y_test, scaler = process_data(file, lag)
minimum = scaler.data_min_
maximum = scaler.data_max_
scale = scaler.scale_
scaled = (maximum * (scale * X_test[0] - minimum * scale))
unscalaed = (scaled + (minimum * scale))/scale
y_test = scaler.inverse_transform(y_test.reshape(-1, 1)).reshape(1, -1)[0]
X_test_2 = scaler.inverse_transform(X_test.reshape(-1, 1)).reshape(1, -1)[0]
print(X_test[0].shape)
with open('scaler_data/{}.txt'.format(model_name), 'w') as output:
output.write('{},{},{}'.format(scale[0],minimum[0],maximum[0]))
# print(y_test.shape)
y_preds = []
for name, model in zip(names, models):
# if name == 'SAEs':
# # X_test = np.reshape(X_test, (X_test.shape[0], X_test.shape[1]))
# else:
# X_test = np.reshape(X_test, (X_test.shape[0], X_test.shape[1], 1))
file = 'images/' + name + '.png'
plot_model(model, to_file=file, show_shapes=True)
predicted = model.predict(X_test)
predicted = scaler.inverse_transform(predicted.reshape(-1, 1)).reshape(1, -1)[0]
y_preds.append(predicted[:384])
print(name)
eva_regress(y_test, predicted)
plot_results(y_test[:384], y_preds, names)
def main():
lstm = []
y_preds = []
for idx in range(4, 13): #training logs from value 4 to 12
lstm.append("lstm_{}.h5".format(idx))
model = load_model('model/{}'.format(lstm[idx - 4]))
file = '970_1_data.csv'
_, _, X_test, y_test, scaler = process_data(file, idx)
y_test = scaler.inverse_transform(y_test.reshape(-1, 1)).reshape(
1, -1)[0] #scaling the model from 0,1 to orignal value
X_test = np.reshape(X_test, (X_test.shape[0], X_test.shape[1], 1))
predicted = model.predict(X_test)
predicted = scaler.inverse_transform(predicted.reshape(-1, 1)).reshape(
1, -1)[0]
y_preds.append(
predicted[:384]) # adding all the lags model prediction to y_preds
print(lstm[idx - 4])
eva_regress(y_test, predicted)
plot_results(y_test[:384], y_preds, lstm)