Add model.parameters.transform() and individual_parameters.transform()

[KaisaridiSofia]

Add functions to ease the passing from the estimated parameters to the ones we actually interpret. This applies to

• the model parameters (fixed effects), where we estimate the $\log(g)$, and $\log(v)$ while we actually need to interpret the $p$ and $v$
• the individual parameters (random effects), where after the personalization we obtain the sources $s$, while we actually need to interpret the space shifts $w$

model = LogisticModel(name="test-model", source_dimension=1, dimension=2, obs_models= 'gaussian-diagonal')
model.fit(leaspy_dataset, "mcmc_saem", n_iter=1000, progress_bar=True, path='my_path')
model.parameters.transform()

this will create a dictionary with

• Population parameters: p, v, betas
  - ++ mixture: probs
  - ++ joint:
• Individual parameters: tau_mean, tau_std, xi_std
  - ++ mixture: xi_mean, sources_mean
  - ++ joint:
• Residual standard error (noise): epsilons

The question is should we include the betas and the sources_mean or should we replace them with the w_mean ?

individual_parameters = model.personalize(leaspy_dataset,'scipy_minimize')
individual_parameters.transform()
``

this will return a dataframe with columns for ID, xi, tau, sources and w