Multivariate Ovulation Window Detection at OvuFriend.

We present new results related to retrospective detection of ovulation days basing on information entered by the users of one of online platforms available in the market. Comparing to our previous studies, we improve the accuracy of algorithms which are based on evaluation and synthesis of multivariate data sources. Results are reported for 224 menstrual cycles which were labeled by medical experts. In the experiments, we pay special attention to the aspect of uncertainty associated with the tagging process.