Package usage comments

[j1c]

Just some thoughts after using the package for a bit now:

• It would be great to have the actual functions in the docstrings for simulations. For example, Function for generating a linear simulation. for linear_sim isn't really useful. You can have the formulation for w vector so that I don't have to look up the actual paper to see the functions.
  • Also allow for custom seeds for simulations.
• (suggestion) it would be great if you followed the sklearn model for writing classes.
  • For example, have the instantiation of the class only deal with the parameters. Currently, instantiating MGC requires the input data, which to me doesn't make much sense.
    • consider using None as default parameter for compute_distance_matrix parameter for MGC.
  • Using a fit function to be the default function name for doing the computation. There are test_statstic and p_value function, and I wasn't sure which one to run at first. Having a fit function eliminates ambiguity.
  • it would be great if you stored the results as class attributes. There was a time when I ran MGC for like an hour and forgot to store the results to a variable, and I had to run it again.
• consider updating __init__.py files for convenience in importing things. For example, i have to type in from mgcpy.independence_tests.mgc.mgc import MGC. It would be nice to do from mgcpy import MGC or something along those lines.

[junhaobearxiong]

• more helpful input error checking
• mgc and dcorr should use the same dist_transform function

[sampan501]

• (suggestion) it would be great if you followed the sklearn model for writing classes.

  • For example, have the instantiation of the class only deal with the parameters. Currently, instantiating MGC requires the input data, which to me doesn't make much sense.

    • consider using None as default parameter for compute_distance_matrix parameter for MGC.

Done

• consider updating __init__.py files for convenience in importing things. For example, i have to type in from mgcpy.independence_tests.mgc.mgc import MGC. It would be nice to do from mgcpy import MGC or something along those lines.

Tried this but had issues with relative imports when using cython