The individual keys of the 3 men The transition concerned are summed to give 1.80. Only the decimals of this sum will be useful later? i.e. 0.80.
The initial (unperturbed) value determines the row of the transition matrix to use (subfigure c) ? here the one corresponding to the value 3. On this row? we are placed at the level of the decimal part of the sum of the keys calculated in the previous step? i.e. 0.80. 5. The perturbation to be applied to the initial value is given by the value of the corresponding
Transition matrix (subfigure c)
(here +2). 6. The perturbed value is then The transition obtained by applying the perturbation to the initial value (here 3 + 2 = 5).
In the distributed data file (subfigure d) ? the perturbed value 5 replaces the initial (unperturbed) value 3.
This perturbation method has many
Although completely random (due to the nigeria phone number library randomness of the keys)? it guarantees that the same perturbation is always applied to the same count regardless of the data table in which it appears. This is because the individual keys are fixed once and for all;
Which guarantee that the statistics ultimately produced do not induce any systematic bias? either downward or upward. In other words? by applying this method the statistics are perturbed downward as often as upward? so that the averages are preserved. This guarantees that in practice? the conclusions drawn from the analysis of perturbed data will be similar 3 steps to identify leads ready to close a deal to those that would have been drawn from unperturbed data;
The choice of the transition matrix
Also makes it possible to guarantee – if desired – that the perturbed values will never take certain values. In the simplified example in Figure 2 ? these are the values 1 and 2: no combination of the initial values aero leads and the perturbations in the matrix can lead to the perturbed value being 1 or 2 (for example? a 1 can only be perturbed with a -1? a +2 or a +3? but not with a 0 or a +1).
