Fit Algorithms

The iterative nonlinear fitting procedure used can be chosen among three methods:

• Least Square (LS)

• Least Absolute Residuals (LAR)

• Bisquare (B2)

For details about the different fitting methods, please consult the white paper produced by National Instruments.

In a nutshell, the LAR and B2 methods are less sensitive to outliers.

Each of these methods can be used with equal weights for all data points (unweighted fit, wi = 1 for all data points) or a statistical weight equal to the inverse of the data value hi (weighted fit, wi = 1/hi, see note at the bottom of this page).

The choice of algorithm and weighting is done in the ALiX Settings >> Histogram Fits tab:

Note: Since most plots are histograms (i.e. curves with integer values), the problem of weighting a data point with value 0 using the formula wi = 1/hi above is solved internally by using a weight of 1 instead. To exclude histogram points with a value of zero, the selected plot area needs to exclude these values.