VoseCalibrate | Vose Software

# VoseCalibrate

VoseCalibrate(DistributionObject, {Data})

• DistributionObject – a continuous distribution object

• {Data} – a set of data values or a data object

The VoseCalibrate function can be used to make a defined distribution match a data set to some degree. Let be the set of k data values, let   be the cumulative probability for x for the defined distribution, and let c be a constant. The VoseCalibrate function returns the constant c such that:

In other words, when the data are multiplied by c their average cumulative probability = 0.5.

Example

DistributionObject = VoseLognormal(2,3)

Data = (2,3,5,6,9}

The data fall at high cumulative probabilities, averaging out to 0.87:

VoseCalibrate(VoseLognormalObject(2,3),{ 2,3,5,6,9}) returns the value 0.252. Multiplying each value by this constant gives {0.50,0.76,1.26,1.51,2.27} which returns an average cumulative probability of 0.5:

The main application of this function is to match a defined distribution (usually by fitting to some much larger set of external data) to a much smaller internal data set. The external data provide the shape of the distribution, and VoseCalibrate provides a constant to better reflect the internal experience. Dividing a sample from the external data-fitted distribution by the value from VoseCalibrate will produce sampled values calibrated to the internal data.

The distribution object input must be a continuous distribution.