Using data from the US Census, this visualization displays popular names that have historically been androgynous.
Control points are not intersected by the path for several non-linear interpolations for lines (“basis” in this case). Because of this, point tracking for non-linear lines should be calculated from the svg-path rather than from the d3-scales. Mike Bostock has two examples showing how to do this with a single path. This gist shows one approach for point tracking with multiple paths. Other approaches are shown here and here.
This approach uses a combination of precomputed voronoi tessellations and on-the-fly closestPoint to determine which line and point to highlight. The voronoi tessellations determine the line, which the on-the-fly closestPoint determines the point. Notedly, the voronoi tessellations are computed using the coordinates producted by the linear scale, rather than the non-linear scale (“basis” here). For this visualization, incorrect line selection does not occur as a result of this under-the-hood inaccuracy. YMMV.