Visualization Optimization and Interpretation of Topic Labels and Color Zones

[Hanqingxu123]

Hello,

First of all, thank you for developing the DataMapPlot library. I'm currently using it and encountered a couple of issues that I hope you can assist me with.

Label placement optimization: In the current visualization, each topic label is connected to the corresponding document region by a line. However, this makes the plot look a bit cluttered. I would like to move all the topic labels to the top-right corner of the plot, similar to how a legend would appear, rather than labeling them directly on the chart. Is there a way to achieve this? If so, how can I implement it?

Interpretation of color zones: In my visualizations, the color zones for each topic appear like "tree rings," forming concentric layers. Could you explain what this effect represents? Is it related to the relevance of documents to the topic, or is it visualizing another dimension of the data?

Thank you again for your help, and I look forward to your response!

Best regards,

[lmcinnes]

There are some options in label placement. See here and here for some of them. None are quite what you are after I believe. In that case you are best just placing labels yourself via matplotlib.

The colour zones are KDEs based on the points in the cluster. You can turn them off entirely (add_glow=False) or control them more directly with glow_keywords . See the end of this documentation for some more details.