Closed wangqianwen0418 closed 3 years ago
In all the three design options, I want to resolve the ambiguity of color encoding.
option 1; For each attribute, we show the attribute distribution of each state. X-axis indicates the value of attributes (works for both categorical and numerical), and y-axis indicates the number of patients.
option 2 simply modify the layout of option 1. So that the layout is consistent with the one used in the sankey diagram (i.e., state in the horizontal direction and attribute in the vertical direction). But the problem is, the reordering makes it hard to compare the distribution of different states.
option 3 is relatively harder to understand than option 1&2. Each patient is represented by a bar whose height indicates the value of the corresponding attribute. The advantage of this design is that a) the encoding can be adopted to the sankey diagram; 2) the width of the chart indicates the number of patients
I have implemented O2 and O3;
-O3:
Great!
I think it would be very helpful if you could move the labels for the variables somehow so that we can see them without tool tips/hovering.
when the height of a chart is too small, add a scaled chart in the tooltip
currently, it is hard to read and compare from the pure-text-based summary.