Hi,
I am new to PyMC3 and I’m currently running my first codes regarding Bayesian Inference.
When I run my pm.Model () code, it returns the plot_traces for my parameters, but without Y-axis values in the “plot_dist” graph in the left area of the plot (namely the “Frequency” values). Both the axis label and the values are missing.
How could I overcome this? Are there any specific directions or lines of code to include in my model in order to appear the y-axis?
I would appreciate it if someone knows what is happening in this theme.
Thank you in advance!
Are you asking about the distributions included as part of plot_trace()? I think the intent of those is to provide information about the sampling procedure. If you are interested in exploring the distributions themselves, you might want to check out plot_dist()?
@cluhmann
Thank you for your answer, I also see this direction.
Indeed, precisely in these distributions I’m referring to.
Navigating to similar topics I see that some of these distributions have “Frequency” and “Sample value” labels as well as visible frequency values in their Y-axes (exactly at the left side of each plot), like the screenshot below:
However, in my case, these values and labels are missing, as you can see in the image below:
I would like to know if there is any differentiation in the code lines in the pm.Model() for extracting the two above plots.
P.S.: I ran some sample code from the Internet in my Jupyter Notebook to see if I get Y-axis labels and values, but even then it doesn’t show them.
I’m wondering if it’s a pymc3 version issue or if I’ve missed the import of a specific library/package that imports the axes and their values.
I can attach part of my code if you think it is necessary.
Strange. Well on the official documentation page for az.plot_trace(), none of the examples have the y-axis labeled. But I’m not sure if that behavior has changed or if it’s an option that one can set. @OriolAbril ?
Old versions (4+ years ago) probably had the axis labels added by default. This is no longer the case. Currently you’d need to manually add them using the array of axes returned by plot_trace.
