The problem is that displayed correlations—for example, between factors and symptoms—may not be very meaningful if, for instance, the factor is only present on a few days. It would therefore be very informative to add a numerical indicator of validity to the percentage value of the correlations. This validity level should at least reflect how many days the measured correlation is based on. Additionally, it could for example also take temporal stability into account or how well the factor can be isolated from other factors in the data—or whether it consistently appears only within a cluster together with other factors. The latter, of course, is statistically more demanding.