Pokin: What are college mathematics good for

We briefly mentioned the new mobile application Pokin, which won the 16th AppParade on our blog a month ago. Now the time has come to look at it in more detail. The attractive exterior hides a relatively sophisticated technology, which we developed in Ackee from scratch and thus became pioneers in this area of ​​its kind.

The application itself is designed to measure sexual activity. Put simply, it is actually a kind of RunKeeper for bed. Pokin can measure the time, number of thrusts, frequency and volume of your partner. You can then compare your performances and see if you are gradually improving in some parameters.

Unlike jogging, where RunKeeper simply tracks your movement via GPS and measures the speed, sex is a much greater challenge from a technological point of view. It is not a problem to collect the necessary data, the difficulties lie in the way of getting the relevant information about the user’s sexual performance. If nothing else, it is because every movement is slightly different and each mattress is a different hardness. The data thus contains plenty of noises that impede direct analysis of sexual performance.

To be able to process the data further, we had to clean it first. This was done with the help of several algorithms such as Savitzky-Golay, which can increase the signal to noise ratio without significant signal distortion. Of the known algorithms we also used Moving Window and Low-passfilter. The actual process of data cleansing is surprisingly very similar to the method for cleaning an analog signal.

Savitzky-Golay filter v akci

At the moment we have cleaned data available, it is possible to embark on their depth analysis. The aim was to find continuity in the numbers and identify relevant sexual performance parameters. Based on the similarities, we were able to determine the series and on the other hand, we were able to reject false positive results. Only at the end of this process did we gain relevant information about the user’s sexual performance that we could then clearly present to them.

This whole process had to be preceded by a thorough mathematical simulation in the Matlab mathematical software, where we first verified our hypothesis, and only then we could transfer it into a code.

We worked on the development of algorithms necessary for data analysis in a team of 4 people and during the course we also subjected certain aspects to consultations with mathematicians. Work on these algorithms took us approximately 80% of the entire application development time, which took about 1.5 months.

The presenting Robert Rosenberg expressed himself regarding user testing at the 16th AppParade, but this is not a topic for this strictly technological blog anymore :-).

At Ackee we are simply not afraid of technological challenges!

Show your support

Clapping shows how much you appreciated Dominik Veselý’s story.