Problems with Cubic Spline in WorkOut 1.5Document ID: DQ051
Relevant to following products: WorkOut 1.5
Document description: I am getting strange results for extrapolated points with the Cubic Spline.
WorkOut 1.5's Cubic Spline reduction averages counts for each standard and uses that for the fit. A problem with WorkOut 1.5 is that when it back fits points which are outside the range of standards (such as some of the replicate points at the low and high standards), WorkOut 1.5 correctly calculates points on the extrapolated Cubic Spline but the nature of the Cubic Spline means that the resulting values are beyond a turning point of the Cubic Spline and do not follow the defined standards data set. Cubic Spline is OK for interpolation (within the defined range of standards) but not recommended for extrapolation. The only solution for WorkOut 1.5 is to use an alternative fit method or avoid extrapolation.
Since WorkOut 1.5 we have had many requests for the ability to safely extrapolate data when using a Cubic Spline. Rather than using the Cubic Spline segments (with continuous derivatives) outside the range of standards (as WorkOut 1.5 does), WorkOut 2.0 supports extrapolation by assuming a linear relationship between X and Y outside of the Cubic Spline range. This can be specified i.e. you can allow or disallow extrapolation with Cubic Spline.