Part of the job mathematics does in science is developing tools for describing patterns - any patterns. We've also learned to deal with uncertainty, so we can now describe approximate, probabilistic patterns as well.

Thanks to this, if there's any pattern about a phenomenon to be found, we likely can describe it already; if not, there's no good reason to believe we won't be able to describe it in the future. That is the first step towards an explanation (second being tying the pattern to theories that fit other observable evidence, and that have predictive power).

If there's absolutely no pattern to be found in a phenomenon, then it implies the phenomenon doesn't have any consequences on anything - which implies it isn't even observable in the first place. From that follows it isn't even worth thinking about.