What does R-sw Discriminant do?
- it allows identifying the most differentiating variables using (1) forward / backward selection or (2) a Genetic Algorithm;
- it provides a set of discriminant coefficients to be used in a typing / allocation tool based on the preferred solution / variables combination;
- it predicts group / cluster membership for a new set of records based on a linear or quadratic discriminant model;
- it allows cross-validation of the model through repeated random sub-sampling validation, to ensure minimal overfitting and thus identify relationships in the data that hold in general.
- in case of forward / backward variables selection, the package produces a full step-by-step report of the selection process, inclusive of the classification accuracy achievable by considering the variables specified; variables are added or removed one at-a-time from the set of variables specified;
- in case a Genetic Algorithm is used for the variables selection, the package produces information on the variables combinations appearing in the final population as well as information on the variables combinations of the population that produced the solution with the highest classification accuracy throughout all the generations (along with the associated classification accuracy);
- variables selection is based on either linear or quadratic discriminant analysis;
- cross-validation through repeated random sub-sampling validation (training and testing model) can be performed at each step to estimate the correct prediction rate of the discriminant model; the number of cross-validation runs at each step is decided by the User (e.g., 100);
- a subset of variables can be forced into the model (fixed variables);
- data can be easily imported and outcomes can be exported as CSV/text files.
Support, Manual and Examples:
- technical support and statistical consulting is available free of charge (within reasonable limits);
- an ‘html’ manual is provided with a detailed description of all available functions;
- full working examples are provided to help the User to become familiar with the package. These examples can be easily adapted by the User for new projects.
Note: this is a pure analysis tool, therefore data must be collected through an external source (such as a CAWI, CAPI or PAPI system).