What do you think of my compromise to deal with the busyness of the jitter plot? Can you think of any other solutions that still result in an attractive, information dense figure?
Heat map. I think each point in this plot is dependent (?), ie samples across taxa sum to 100. A heat map would also allow grouping eg by kmeans or other distance, and patterns can be discerned much more easily (imho). Related, I think heatmaps are generally under-utilised in microbial ecology and stacked barcharts overused.
Hey Olin - thanks! I looked at heatmaps and wasn’t totally sold. I definitely need to come back to add clustering. I wonder if the benefit of clusters is for more samples/taxa at which point there are too many to actually get anything out of the visual. Here’s the heat map episode ruclips.net/video/-4sGamyqGAQ/видео.html
What do you think of my compromise to deal with the busyness of the jitter plot? Can you think of any other solutions that still result in an attractive, information dense figure?
Heat map. I think each point in this plot is dependent (?), ie samples across taxa sum to 100. A heat map would also allow grouping eg by kmeans or other distance, and patterns can be discerned much more easily (imho). Related, I think heatmaps are generally under-utilised in microbial ecology and stacked barcharts overused.
Hey Olin - thanks! I looked at heatmaps and wasn’t totally sold. I definitely need to come back to add clustering. I wonder if the benefit of clusters is for more samples/taxa at which point there are too many to actually get anything out of the visual. Here’s the heat map episode ruclips.net/video/-4sGamyqGAQ/видео.html
Thanks
My geom_boxplot doesn't show the individual data points, any explanation as to why?