Best way to extract parameters and save model objects?

[hauselin]

Hi, I'm wondering are there any recommended workflows/guidelines for extracting parameters from fitted objects and saving model objects? I can't find any guidelines in the documentation.

It seems like there are many ways to extract parameters from fitted objects: model.get_dependence(), model.get_model_parameters(), model.dependencies. Which way is preferred? (I'm planning to extract and store the parameter values in dictionaries/dataframes so I can run follow-up analyses.)

Does PyDDM have built-in functions/methods for saving objects? I've been using np.savez() to save fitted objects to disk and np.load() to reload them from disk. Is this the best way to save objects?

Really appreciate any help anyone can provide. Thanks!

[mwshinn]

Thanks for your inquiry!

I would not recommend using get_dependence() or model.dependencies, since these will make it difficult to extract the information you need.

If you want to save a model to reuse later, the easiest (and most space-efficient) way to do this is to print() the model, cast it to a string with str(), or run repr() on it, and then save the string they output. The function print() produces a nicer formatting than repr(), but otherwise they are identical. (str(), print(), and repr() I am talking about are all the versions built-in to Python) If you execute these strings on a new python interpreter with the appropriate imports, you should get the same model.

For analyzing parameters, get_model_paramters() gives you what you want, and get_model_parameter_names() gives you the name. They are guaranteed to be in the same order, so you can use these lists to easily construct a dataframe.

I just updated the quickstart guide in the documentation to clarify all of this!

[hauselin]

Perfect! Everything worked and thanks for updating the documentation. Thanks again, Max!