Introduction into Calibration with panorama
This guide is designed to give a short overview of setting up a new calibration model. It is not an introduction into Chemometrics nor does it discuss in detail the effect of special parameters, influencing the overall calibration module.
To start a new calibration you need a panorama project containing your own sample files or use the existing project “CowMilkRange.project” from the demo data coming with the installation.
Load the project of interest and make sure it is selected in the project explorer.
From the Quantify menu select New Multivariate Calibration. A wizard opens, which guides you through the eight steps of setting up a calibration model.
Step 1: Calibration Wizard Steps Overview
The calibration steps overview page lists all relevant steps for a multivariate calibration. No user interaction is required during this step.
Step 2: Entering General Information
Step 2 is meant to enter a descriptive name for the calibration model. This name is used later on to identify the calibration model within the project. Optionally, you might give some more detailed description to the purpose of calibration.
Step 3: Calibration Model and appropriate Parameters
Step 3 plays a key role. At this point you need to define on which kind of model the calibration will be calculated, e.g:
In addition to the calibration model some specific method parameters e.g. for the automatic outlier detection can be customized.
Quantization using statistical evaluation does not work without numeric sample property values, e.g. concentrations. Such values will be stored in so called labels of loaded spectral data. A list of all applicable labels is shown to select the appropriate sample property to be calibrated. In this example the “Pro” label is used.
New labels can be created and filled easily as well.
Step 4: Spectra Selection
After definition of a calibration model it is important to select a set of appropriate calibration spectra. Here also a set of independent validation spectra might be assigned. All spectra in the project are available here.
For convenience the actual calibration spectra selection is shown in the upper spectral area together with some statistical information like the correlation (green line), average (red line), variance (light red shape) etc.
The table also lists the numeric contents of the calibrated label which has been selected previously. In this example the “Pro” label is used.
Step 5: Preprocessing - Applying mathematical Operations
In particular cases it might be necessary to preprocess spectral data, e.g. normalize or calculate a derivative. All applicable mathematical operations provided by the software are available here.
Step 6: Definition of relevant spectral Ranges (Variable Selection)
Selection of significant variables for calibration is another important step. You can choose the variables as spectral ranges conveniently. Spectral ranges can be adjusted easily either graphically with drag and drop or numerically in the table.
Statistical information like the correlation (green line) and variance (light red shape) and some advice from the software will help you to make your decision.
Final Step: Calculated Calibration Model – Ready for Review and interactive Optimization
The final step presents a summary of evaluation results on multiple screens. The example shows the result of a MLR Calibration report displaying the settings, basic and derived data like regression, calibration results and outlier statistics.
Prediction Plot
The prediction plot shows predicted values derived from the calibration model versus actual values of the data object. Each square in the plot represents a calibration data object. Outliers are easily identified as red colored squares.
Residual Plot
The 2D residual plot is useful to detect outlying samples in the calibration set from the residual.
Prediction plot and residual plot are examples for numerous plots available in the software, like 2D and 3D Scores Plots, Loadings Plot, etc.
Final Overview
The overview shows a summary of derived calibration results added to the list of spectral data similar to the view in step 4.
From any of the plots, the overview and parameter settings in previous steps the calibration model will be refined by subsequent parameter change and recalculation of the calibration model.
After completion of the optimization, results can be printed and the calibration model is saved.
Nevertheless the data can be modified at a later stage using the Edit mode within the project.
For a more detailed introduction with examples
>>> CLICK HERE <<<