2.62 Rank Models by Fit Reports

Summary

This Rank Models by Fit Reports app can be used to rank fitting functions by fit reports. It can find the optimal model by AIC, BIC, Adj. R-Square, Reduced Chi-Sqr or Residual Sum of Squares.

Tutorial

Prepare Fit Reports

  1. With a worksheet window activated, select Data: Connect to File: Text/CSV in the menu to import the sample file <Origin program folder>\Samples\Samples\Curve Fitting\Exponential Decay.dat.
  2. Highlight Column A and B, choose Analysis: Fitting: Nonlinear Curve Fit menu to open NLFit dialog. Select ExpDec1 as Function, and fit the dataset.
    Rank Models by Fit Reports 01.png
  3. The Fit Report is output in the FitNL1 sheet.
  4. Follow above steps, fit Column A and B dataset by ExpDec2 and ExpDec3 function.
  5. Now, we have three fit reports by three functions for the same source data.
    Rank Models by Fit Reports 02.png

Rank Models by Fit Reports

Here, show how to use Rank Models by Fit Reports app to find the best fitting function according to given fit reports.

  1. Click Rank Models by Fit Reports icon Rank Models by Fit Reports icon.png in Apps Gallery window.
  2. In the opened dialog, choose 3 for Number of Models to Compare. Click OK button.
    Rank Models by Fit Reports 03.png
  3. In the Choose Fit Reports dialog, select three fit reports. Click OK button.
    Rank Models by Fit Reports 04.png
  4. A summary worksheet will be shown. It is sorted by BIC. BIC result shows ExpDec2 function is the best model. The output sheet also shows rank results according to other methods, e.g. AI, Adj. R-Square, Reduced Chi-Sqr and Residual Sum of Squares. The optimal models determined by these methods are marked in a red color. In this example, all methods show that ExpDec2 function is the optimal model.
    Rank Models by Fit Reports 05.png