Can it save combination of ADs as a model ? about adtk HOT 3 OPEN

Yes, you may save your model object by pickling it, something like the following

my_model = MyModel(my_param)
my_model.fit(my_training_data)

with open(my_pickle_path, 'wb') as f:
    pickle.dump(my_model, f)

And when you need to reuse it,

with open(my_pickle_path, 'rb') as f:
    my_model = pickle.load(f)

my_model.detect(my_testing_data)

from adtk.

@tailaiw Pls let me know

from adtk.

@tailaiw, when the return type of fit(...) function is None, how do you save/load it by pickling and then run detect(...). Doesn't it return the following error? (I am referring to AutoregressionAD)

'NoneType' object has no attribute 'detect'

P.S. I appreciate that you discuss at this link to some extend, but it is confusing having such function (fit) but not able to use it as "fit and predict" fashion in terms of API design.

from adtk.