A main difficulty in using these programs is finding appropriate starting values. This is very important because these likelihoods are multi-modal and minor modes can easily be produced when starting values are inappropriate. In addition, default starting values are poorly chosen for most programs, so intelligent user selected starting values are best.
This R program allows you to visually select starting values. All you need to do is provide a "csv" (comma separated values) data file in any format, and the program will interactively present you with plotted curves for your subjects (in a random order). You just click to place each curve in a group based on your judgement. You can choose the number of groups as you go along.
After categorizing some or all of your data, the program will calculate reasonable starting values.
All links active 12/3/2008. Report broken links to
Back to my Home Page