Mark Excluded Points for Refit¶
When exploring a model that is appropriate for a given data set, outliers may be apparent. When this occurs, it is often desirable to immediately tag the outliers and refit the data excluding these points. The Mark Excluded Points for Refit option is used to automatically select outliers or points where math errors occur. The outliers can also be tagged manually.
Manually Marking Outliers¶
The Mark Excluded Points for Refit option is not needed to manually select outliers. To mark a point for exclusion in a refitting procedure, left click on the point desired. The point will be redrawn with a surrounding circle. To unmark a point that has been marked, click once again with the left mouse button. The surrounding circle will be removed.
Although this option is generally used to mark and remove outliers that were previously fitted, it can be used to mark inactive points that were excluded from the original fitting. In this instance, a refit will activate the previously excluded points.
Automatically Marking Outliers¶
The Mark Excluded Points for Refit option offers the following means for automatically marking outliers:
- None (Clear All Marked Points) - All marked points are cleared. The current active and inactive state of the points is locked.
- All Points with Math Errors
- All Points Outside 2, 3, or 4 Standard Errors
- All Points Outside 90, 95, 99, 99.9% Prediction Interval
Refitting¶
If one or more points is currently marked, one of the refit buttons will be active within the control panel of the Curve-Fit graph. This button is adjacent the Mark Excluded Points for Refit button. When a refit is initiated, all marked points are toggled in state and a full refit is made. Marked active points are made inactive while marked inactive points are made active. It is thus possible to go back and forth between including and excluding a set of marked points by successive refits.
Note that a refit may result in the previously selected equation not being available. There may also be more or fewer equations in the list after a refit. When points are toggled off, a rational function that was previously defined fully across the X-range may have undefined regions or poles in the vicinity of the points toggled off. The equation may still be available if the Rational Defined Xmin to Xmax filter is disabled in the Equation List, but the new fit may be of dubious value.
If a refit fails completely so that no equation is available, the original state of the data is restored.
A refit completely processes all equations previously fitted. This may require some time with large data sets, depending on the non-linear equations selected.