You might consider bypassing CART and trying other, more direct, "brute-force" approaches.
If the diagnostic tests all have binary outcomes, you can use a learning dataset to test all possible sequences of tests, and rank sequences according to utility or cost-effectiveness. CART is not guaranteed to produce the optimal solution. A brute-force approach can guarantee the optimal solution. Another advantage of a brute force approach is that you can consider collateral information, such as costs of tests, ease of administration, or logical priority. There are also some Bayesian models which might be better than CART for selecting a good test sequence. -------------------------------------------------------------------------------- John Uebersax, PhD (858) 597-5571 La Jolla, California (858) 625-0155 (fax) email: [EMAIL PROTECTED] Statistics: http://ourworld.compuserve.com/homepages/jsuebersax/agree.htm Psychology: http://members.aol.com/spiritualpsych -------------------------------------------------------------------------------- . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
