You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I will start working on this. Here is the main idea to make sure we are on the same page.
def plot_compare_pointwise(idata_dict, ic="waic", color=None, ...):
"""Graphical comparison of multiple models in terms of pointwise ELPD.
idata_dict : mapping (str -> InferenceData)
Dictionary mapping the name of each model to compare to its corresponding InfereceData
ic : str, optional
Information Criterion (WAIC or LOO) used to compare models. Default WAIC.
color : str or array_like, optional
Colors of the scatter plot, if color is a str all dots will have the same color, if it is the size of the observations, each dot will have the specified color, otherwise, it will be interpreted as a list of the dims to be used for the color code.
"""
# behaviour similar to plot_pair with pointwise ELPD plots
I was also thinking that in addition to the color code, it may be useful to have a coords argument, in order to plot only a subset. It could be interesting in the radon case for instance. I will give some more thinking to the API and start a PR soon.
ArviZ is missing function to compare models in terms of pointwise ELPD. See figure 10.
Such a function could work for 2 models (like figure 10) or for more than 2 and return a grid of non-redundant subplots, like
plot_pair
.The text was updated successfully, but these errors were encountered: