estimate
Uses the method described on page 360 of Law (2007) Simulation Modeling and Analysis, ISBN 0073294411, 9780073294414 There must be at least three observations within the data and there must be at least three different values. That is, all the values must not be the same. The observations should all be greater than or equal to 0.0. That is, no negative values are allowed within the data. If these conditions do not hold, then 0.0 is returned for the shift. That is, no shift.
The value tolerance is used to consider whether the computed shift value is close enough to zero to consider the shift 0.0. The default is 0.001. That is, a value of the estimated shift that is less than the tolerance are considered 0.0.
This approach estimates a shift parameter that is intended to shift the distribution to the left. If X(i) is the original datum, then the shifted data is intended to be Y(i) = X(i) - shift. Thus, the distribution is shifted to the left. The estimated shift should be a positive quantity.