Configure Multivariate Simulation Transformation Properties
The Transformation tab allows you to specify the multivariate transformation to convert the input drill hole data into independent Gaussian factors.
Activity Steps
- Select the Multivariate Simulation in the Project Tree.
- Select the Transformation tab in the Property Panel.
- Configure the following options as required.
- Assays field group:
- Assays To Transform – Select assays to use during the simulation.
- Use compositional transformation – Perform a compositional transformation prior to the Multivariate Normal Transformation into independent Gaussian factors.
- Type of transformation – Select from one of the following options.
- Centre log-ratio (CLR)
- Additive log-ratio (ALR)
- Isometric log-ratio (ILR)
- Compositional value – Value of the complete composition (For example, '100' for %, or '1000000' for ppm).
- Denominator – Select a denominator from any of your assays or a filler option.
- Use Filler – If your raw variables do not add up to the complete composition, add a filler variable during the transformation, which makes up the difference.
- Multivariate Normal Transformation field group:
- PPMT – Projection Pursuit Multivariate Transformation. Only one of the PPMT and GPCA fields can be checked.
- Min Iterations – Minimum number of iterations to use during the PPMT process.
- Max Iterations
- Target Gaussian Percentile – Gaussian percentile at which the PPMT process stops if the percentile is reached.
- Use Spatial Decorrelation – Perform an additional decorrelation process at a specified lag using Min/Max Autocorrelation Factors (MAF).
- Lag Distance – Lag distance for use in the MAF process.
- Lag Tolerance – Equal to half the Lag Distance. Read-Only.
- GPCA – Gaussianisation using Principal Component Analysis.
- Iterations – Number of iterations to perform during the GPCA transformation. Only one of the PPMT and GPCA fields can be checked.
- Assays field group:
- Select Transform to run the transformation.
A set of continuity analyses for each factor display in the Project Tree, as well as a folder with the multivariate transformation name containing all the scatter plots between factors.