def trainModel(unitsInHiddenLayer=8, debug=True):
if debug:
print("Reading Training Data MLP Model (Currency Exchange problem)")
# Reading Training data
trainData = utl.readDataSetAsMatrix(\
utl.CURRENCY_EXCHANGE_TRAINING_FILE(\
utl.SAMPLING_TYPE.HOURLY
)
)
width = utl.HOURLY_WIDTH
inputs = utl.createPattern(trainData[:-1, 2:], width)
outputs = trainData[(width + 1):, 2:]
if debug:
print("Training Model MLP Model (Currency Exchange problem)")
configuration = utl.MLPTrainProcessConfiguration()
configuration.unitsInHiddenLayer = unitsInHiddenLayer
configuration.outputLayer = LinearLayer
configuration.maxEpochs = 1000
configuration.learningrate = 0.001
return utl.trainMLPNetwork(\
inputs,
outputs,
configuration
)
def trainModel(unitsInHiddenLayer = 8, debug = True):
if debug:
print("Reading Training Data RNN Jordan Model (Currency Exchange problem)")
# Reading Training data
trainData = utl.readDataSetAsMatrix(\
utl.CURRENCY_EXCHANGE_TRAINING_FILE(\
utl.SAMPLING_TYPE.HOURLY
)
)
inputs = trainData[:, 2:]
if debug:
print("Training Model RNN hourly Model (Currency Exchange problem)")
configuration = utl.RecurrentTrainProcessConfiguration()
configuration.unitsInHiddenLayer = unitsInHiddenLayer
configuration.maxEpochs = 100
configuration.learningrate = 0.001
configuration.momentum = 0.95
return utl.trainJordanRecurrentNetwork(\
inputs, 1, configuration
)
def trainModel(unitsInHiddenLayer = -1, debug = True):
if debug:
print("Reading Training Data RBF Model (Currency Exchange problem)")
# Reading Training data
trainData = utl.readDataSetAsMatrix(\
utl.CURRENCY_EXCHANGE_TRAINING_FILE(\
utl.SAMPLING_TYPE.HOURLY
)
)
width = utl.HOURLY_WIDTH
inputs = utl.createPattern(trainData[:-1, 2:], width)
outputs = trainData[(width+1):, 2:].astype(np.float64)
if debug:
print("Training Model RBF Model (Currency Exchange problem)")
configuration = utl.RBFTrainProcessConfiguration()
configuration.outputLayer = LinearLayer
if unitsInHiddenLayer <= 0:
configuration.performClustering = False
else:
configuration.unitsInHiddenLayer = unitsInHiddenLayer
configuration.performClustering = True
configuration.maxEpochs = 1000
configuration.learningrate = 0.001
return utl.trainRBFNetwork(\
inputs,
outputs,
configuration
)