![]() |
Probability feedback loops |
![]() ![]() ![]() |
When an activity feeds back to a previous activity, a recursive feedback loop is created and a probability that this feedback loop occurs must be defined.
Note: A start and end box must be present when working with feedback calculations and they must exist outside of the feedback loop.
Example
The probability of the first activity (Start) occurring is p(Start)=100%.
The probability of Step 2 occurring is p(Step 2)=100%.
Therefore, the probability of Step 3 occurring is 100% x 50%=50%.
Now, there is a 10% chance that Step 3 feeds back to Step 2. Therefore the chance of Step 2 occurring a second time is 50% x 10%=5%.
So the effects of this feedback loop are as follows:
The probability of Step 2 occurring remains at 100%, although there is an affect on the volume of transactions. In this example, there is a 5% chance that Step 2 will need to be repeated. This means that the volume of Step 2 of one pass through the feedback loop is 105% of the original volume. Simply put, 100 forms may be passed to Step 2, 5 may have to be redone, therefore the volume becomes 105 forms in total, accounting for the probability of one loop back to Step 2.
If the loop is encountered a second time, i.e. Step 2 is repeated a third time, the probability is 5% x 50% x 10% = 0.25%. The chance that it feedbacks again and Step 2 is repeated a fourth time is 0.25% x 50% x 10% = 0.0125%, fifth time is 0.0125% x 50% x 10% = 0.000625%.
So we now have a series of probabilities that Step 2 occurs:
One = 100% Two = 5% Three = 0.25% Four = 0.0125% Five = 0.000625%
If we set the number of feedback iterations at 5 to cover the chance that one of the above situations will arise, the recursive feedback factor is the sum of all probabilities.
Feedback factor = 105.26%
Note: By default, TIBCO Nimbus® automatically sets the number of iterations at 20.
|