# Start Testing
# Create an instance of the filter
kf = Kalman(history, order)
# Use the first #history of them for training
obs_train = obs[:history]
model_train = model[:history]
# The rest to be used dynamically
obs_dyn = obs[history:]
model_dyn = model[history:]
fcst = np.zeros_like(obs_dyn)
# Perform an initial training of the model
kf.trainMe(obs_train, model_train)
for ij in range(len(obs_dyn)):
# Provide a correction to the forecast
fcst[ij] = kf.adjustForecast(model_dyn[ij])
# Update filter
kf.trainMe([obs_dyn[ij]], [model_dyn[ij]])
fig = plt.figure(figsize=figSize)
plt.plot(obs_dyn[pltRange], label='obs')
plt.plot(model_dyn[pltRange], '--', label='model')
plt.plot(fcst[pltRange], '*', label='kalman')
plt.legend()
plt.show()
class HybridKB(object):
def __init__(self, historyKal, dimKal, historyBayes, distType=None):
# First check Kalman related parameters
if historyKal < 1 or dimKal < 1:
print('Hybrid module reporting: ')
raise ValueError(
'Improper value entered for history length and/or dimension of observation matrix (Kalman)...'
)
if dimKal < 2:
print('Hybrid module reporting: ')
print(
'Caution! Low accuracy due to the size of the observation matrix. (Affecting mostly Kalman)'
)
# Now check for Bayes related parameters
if historyBayes < 1:
print('Hybrid module reporting: ')
raise ValueError('History length is too small (Bayes)...')
# Final check to ensure smooth operations
if historyBayes < historyKal:
print('Hybrid module reporting: ')
raise ValueError(
'History length for Bayesian system must be greater or equal to that of Kalman. '
)
# Check if the distribution is supported
if distType is None:
correctionType = 'none' # Initialise type of data to be handled/corrected (will be detected)
else:
if distType.lower() in distAllowed.keys():
correctionType = distAllowed[distType.lower(
)] # Initialise type of data to be handled/corrected (user defined)
else:
print('Hybrid module reporting: ')
raise ValueError(
'Distribution not currently implemented for the system.')
# Now ready to use the input (could defer to classes but thought better to do it here as well)
self.KF = Kalman(historyKal, dimKal)
self.BS = Bayes(historyBayes, correctionType)
def trainMe(self, obs, model, retarget=False):
"""
Master method to control the
initial training of the system.
"""
# Ensure it's working with numpy arrays
myObs = np.array(obs)
myModel = np.array(model)
# Check if shapes match
if myObs.shape != myModel.shape:
raise TypeError('Training set does not have conforming shapes.')
# Train each module
self.KF.trainMe(myObs, myModel)
self.BS.trainMe(myObs, myModel, retarget=retarget)
def adjustForecast(self, model, buff=20.0):
"""
Method to control the correction
of the forecast value.
"""
tempVal = self.KF.adjustForecast(model, buff=buff)
ret = self.BS.adjustForecast(tempVal, buff=buff)
return ret
