Here is an overview of the work done so far relating to plotting:

• Add calibration curve to plotting module #8425 - Decision Regions of Classifiers
• [MRG] Add plotting module with heatmaps for confusion matrix and grid search results #9173 - Confusion matrix and grid search
• Add calibration curve to plotting module #8425 - Calibration curve
• [MRG] Partial dependence plots -- continued #12599 - Partial Dependence

To help control the scope of sklearn.plot, I propose we only do plotting on the axes level and not the figure level. The user would pass in the axes as a keyword. As a convenience, the default of axes will be None. Only in this case, will the plotting function generate a axes/figure to plot on.