python - Discrepancy of HMM model outcome between different process methods

Question

I am using HMMlearn model to analysis over 100 csv files. I first concatenated all csv file to a big one and inputed to HMM model and got good outcome. Then, I tired to used the same HMM model to process each csv file separately and then concatenated the outcomes lastly. I found the outcome from separately processing is different from the one from processing the big csv file.

Does anybody know what cause this discrepancy?

model = GaussianHMM(n_components=5, n_iter=100, covariance_type="diag", verbose=True, random_state=0)

Thank you so much in advance!

Answer 1

等待大神解答