Counters record tolerance of adverse attributes

Counters record tolerance of adverse attributes

Previous Top Next

The counter matrix represents the extend of tolerance of non-ideal values. This is done from the perspective of the current acceptable example that happens to be candidate. Low values like 0.0, 0.1 and so on almost always and in every factor find confirmation. For the values of relevance it is crucial to find the highest confirmable tolerance value in every dimension. In every dimension, the value 0.9 may be confirmed if all other dimensions show the tolerance value of 0.0 at that time. That is not the desired result, as the highest value of each factor is searched for that is compatible with the highest possible values in all other factors.
The counters indicate the performance of each tolerance option between 0 and 1. They represent how often a tolerance option has been successfully tested in former test cycles.
Those counter values map properties of the example base and the current candidate.
The results of tests of different tolerance options are registered, transformed to new counter values and the numbers of successful tests later compared. To each tolerance option in each dimension belongs one counter value. This value records the success of former test of tolerance values. A tolerance option that has had many successful test is represented by a high counter value. Each successful test enlarges this value up to a certain limit. Conversely, unsuccessful tests of tolerance options lead to a reduced counter value. Counter values are subtracted if a tolerance hypothesis proves inadequate when comparing the candidate with the rest of the examples.
At least the counter value 1 remains. That is because no tolerance option can be totally excluded as being totally inadequate. If a tolerance option is represented by the value of 1 this may be interpreted that this tolerance option has not been tested yet or previous tests have failed. The upper limit of the counter values makes sense as the proportion of all counter values represents the probability of a tolerance option to be selected as a tolerance hypothesis in future test cycles. If single counter values would increase too much, these values would be selected preferably and therefore prevent other tolerance options to be selected as tolerance hypothesis.
During the process of analysis the procedure enlarges and reduces the counter values. The proportion of the different counter values represents, how tolerable a specific value (namely the one of the current candidate) in the respective dimension appears.

High counter values indicate a well established evaluation of the respective dimension.

If a high value of tolerance appears well established, the respective dimension appears tolerable and little important.

The counter values represent properties of the relation between the candidate and the reference examples and the target criterion. The counter matrix does not record successfully tested tolerance test vectors but meta information on them. That reduces the information load dramatically.