MOPED

Download Version 3.2 updated 13 April 2020

MOPED was developed to make life easier for people who carry out multivariate analyses by providing a series of linked procedures to:

• display patterns and cluster similar data together, including TWINSPAN, SOMs and UPGMA
• develop models:
  • Discriminant ,
  • Linear regression (least squares and least absolute deviation),
  • Loess
  • Generalised additive (GAM normal, binomial, poisson, gamma)
  • and use the models for prediction.
• Scatterplots with numeric or data/time data and curves plotted through data using:
  • Linear
  • Log-Log
  • Lowess
  • Loess
  • GAM
  • Quantile (linear)
  • or user supplied function

Variables can be included in models as polynomials, forward or all subset stepwise techniques

can be used to select the "best" set of variables.

The steps in a typical analysis are:

• open Data file of species and environmental data
• cluster analysis of species using one of four procedures
• review clusters and merge if required
• discriminant, regression, or GAMs analysis to relate clusters to environmental data
• check model by using the original data file to predict species from environmental data
• save model for future use.

The model can then be used by entering environmental data for a site (or reading it from a file) and predicting species abundance (and likelihood of occurrence for discriminant or GAM logistic models) at the site.