Daily Archives

One Article

Posted by knomee on

Large performance improvements for new Knomee release !


This month's release for Knomee (v2.04) turns out to be quite important, because we were successful to optimize and to reduce Knomee memory consumption significantly. Since iOS kills apps that consume the most memory when it needs to load something new, this behavior (Knomee stopping with no warning because OS killed it) has started to happen before. The improved algorithm is both much faster and less memory-hungry, by a factor of five. This translates in better stability and performance.

As with the previous releases of Knomee 2, this release comes with a number of bug fixes and minor usability improvements. In addition, we have made two changes to reflect some of the comments sent by our users :

  • The main screen, that is used to enter new self-tracking measures, has been using “mountains” representations to give you a quick feedback on how you are doing, with respect to the target that you have set, with each of your quest trackers. Rounded light-green mountains are indicators of success (target reached) whereas rugged dark mountains indicate that you are still far from your target values. To make this more obvious, we have just added a smiley face, happy or unhappy, to the mountain icons. In addition, since 2.04, you may click on the mountain to get an explanation. The weather illustration gives you a trend: a bright sun indicates improvement while rain represents a decline in performance.
  • The Quest Rating interface, that you see when you click on the quest score on the main screen if you have selected the “tech mode” has been simplified. The goal of this interface is both to explain the quest score (from 0% to 100%) and to show the structure of the quest with colors to help you see which trackers help or do not help towards your quest. The new interface is simpler, it show that the score is the sum of three components:  the data score (how much data you have collected), the pattern score (the capability of knomee to extract patterns from your data) and the forecasting score (how well Knomee is doing with its forecasting algorithm). The idea, inspired by Granger causality, is that a good forecasting score is an indication of a good quest (remember that a quest is a "causal hypothesis" that you are trying to prove or disprove).

Knomee has now reached a good usability level and our goal is to develop our reach. There are a number of new competitors to Knomee, which is a good thing, since it shows the need for a truly customizable self-tracking app. However, Knomee is still the only mobile application with embedded AI that gives smart feedbacks and visual analysis of your self-tracking data.