pandas - Python: Automatically choose parameters for ARMA model -
i trying fit arma model time series data. haven't find functions can automatically choose parameter. below code wrote beginner python hence believe code can optimised.
can give me ideas on how to:
- do vectorization on double loop
- quicker way parameter choosing
much appreciate.
parameter_bound = 3 # creating 2-d array, storing residuals of 2 different parameters of arma model residuals = [[0 x in range(parameter_bound)] x in range(parameter_bound)] model = [[0 x in range(parameter_bound)] x in range(parameter_bound)] # calculate residuals each parameter combinations in range(parameter_bound): j in range(parameter_bound): model[i][j] = sm.tsa.arma(input_data, (i,j)).fit() residuals[i][j] = sum(abs(model[i][j].resid)) # find parameters lowest residuals parameters = np.argmin(residuals) parameter1 = parameters/parameter_bound parameter2 = parameters - parameters/parameter_bound*parameter_bound # use model lowest residuals prediction data prediction = model[parameter1][parameter2].resid + input_data
i'm not sure you're expecting, replace lists numpy arrays (i don't think it'll improve specific code):
import numpy np residuals = np.zeros((parameter_bound, parameter_bound)) model = np.zeros((parameter_bound, parameter_bound), np.object)
also, aware np.argmin axis=none returns index flattened array, if want return model parameters of model lowest residuals might try:
prediction = model.ravel()[np.argmin(residuals)].resid + input_data
Comments
Post a Comment