Added value with data mining
At the end of a quality check with the KPS, one of two answers is possible: either the tested telematics system, its wiring, and its signal path are in order, or there is a problem. The KPS actually poses a large number of questions that ultimately lead to this dichotomous answer, which is critical for quality assurance. The information obtained "en route" is therefore anything but junk data. With computational search techniques like data mining, valuable knowledge can be "distilled" out of the huge stockpile of data built up by the KPS from day to day.
An example shows how: Due to the introduction of a new device, certain types of breakdown could occur more often. Data mining can be used to clarify whether all the occurrences of failures have things in common. A typical result might be as follows: the digital detective work shows that these errors occur almost exclusively in cars for which the customers have ordered the same combination of two or more pieces of optional equipment. An investigation of the equipment itself then reveals, for example, that drawing in the cable for equipment A can easily loosen a plug connection for device B. With KPS data and the right mining methods, this can easily be prevented in the future — before the customer experiences the mishap.
To prevent the latter from happening, the KPS makes use of an additional strength, its high testing accuracy. In devices that operate with high-frequency technology, the signal strengths must be determined as precisely as possible. If the plug connection between the GSM antenna and cell phone case is only a little loose, or if the plug is disconnected but right next to the antenna socket, the signal strength may drop only a few decibels. However, when the driver drives over a pothole, the plug may drop off altogether or move farther from the socket. As a result, the cell phone no longer has any reception. However, because the KPS senses a decrease of as little as 5 dB, it can reliably reveal installation defects of this kind that would otherwise be difficult to detect.