Python visuals are inconsistent in the Power BI Service (user-facing web interface) vs Power BI Desktop

Python visuals are inconsistent in the Power BI Service (user-facing web interface) vs Power BI Desktop. I imagine updating packages will help with this.

For example,

•Python code results in unpredictable errors (e.g axes.get_xticks())

•Matplotlib method in Power BI Service returned NULL instead of expected data

•Legend elements are not in the order in which their corresponding plots were added to the graph

•A solution different from standard Matplotlib documentation is required to allow the correct alignment and behavior of legend, which is an unstable and unsustainable solution.