The reason is that Crossfilter is designed to filter consistent views. In this case, you usually filter the visualized dimension and want other dimensions to vary depending on your filter. But the size in which the filter is defined must remain constant, partly because it will be redundant (the filter mechanism is usually displayed visually already), and partly because you do not want your size values ββto jump when you try to filter them.
In any case, in order to get around this, you can define two identical sizes by the same attribute. Use one dimension to filter and the other to group. Thus, with Crossfilter, your filtering dimension and grouping dimensions are separate.
Ethan jewett
source share