Choose Language Hide Translation Bar

Building Predictive Models for Spectral Data

 

See how to:

  • Build a sustainable empirical model based on spectral data/wavelengths
  • Examine data using Graph Builder to get idea of what different spectra look like
  • Use Multivariate Analysis to examine all wave lengths and resulting Correlation Coefficients to confirm multicolinearity
  • Use Model-Driven Multivariate Control Charts to examine all wave lengths variables, drill into runs that are out of control and identify how many Principal Components you need to build model
  • Take spectra of the desired samples, understanding that there is no need for output (Y) information at this point
  • Identify most prominent dimensions in the spectra by Row with Functional Principal Components from Functional Data Explorer
    • Use Functional Principal Component Profiler to get an idea how your spectra are changing as variable of interest changes
    • Save Functional Principal Component scores to data table to use in experimental design
  • Create an experimental design using the functional Principal Components as factors (covariates)
  • Run the experiment to gather the output of interest
  • Model the results via PLS, and/or Generalized Regression (or other methods able to handle correlated factors)
  • Determine the overall optimum solution
  • Use this sustainable model to determine the outcome for all future samples that are analyzed using the same calibrated instrument
  • Use Score plot to examine Categorical Data

Note: Q&A is included at times 24:21, 25:01, 37:01, 37:49, 38:34 and 42:08.