Hi,
really hope that anybody can help. How to consider e.g. different usage frequencies in SPSS TURF? Lets say I have 4 different products (P1 to P4), 5 respondenst and 4 different possible usage frequenies. How to consider rel usage frequenies in SPSS and not only reach? Also: What's wrong with the Syntax. It handles CRITERIA as if 1 ,2 ,3 are the same as "1". Many thank in advance! values 0 "no usage" 1 "usage 1 time per week" 2 "usage 2 times per week" 3 "usage 3 times per week". data p1 P2 P3 P4 1 1 3 3 2 1 3 3 3 1 3 0 0 0 0 0 0 0 1 0 Syntax: SPSSINC TURF VARIABLES = P1 P2 P3 P4 /OPTIONS BESTN = 5 NUMBERTODISPLAY = 2 THRESHOLD = 0 CRITERIA = 1 2 3 REMOVALCYCLES = 0. MAXSETS=5000000 /IMPORTANCE STRICT=YES. 
TURF is designed for ratings, not frequencies. The logic is to find bundles of products that maximize the probability that a customer will like at least one item. Also, I hope your actual data is much larger than what you posted. The CRITERIA keyword takes a list of values that are considered positive responses, so CRITERIA = 1 2 3 means that all of those are positive responses. It does not distinguish among them. This is consistent with the probability statement above. In effect, in your case, you are looking for the bundle with the most positive value as they all indicate a favorable response. HTH, On Thu, Jan 26, 2017 at 8:11 AM, nocheinschmidt <[hidden email]> wrote: Hi, 
Hi Jon, many thanks for that quick response! Just started dealing with TUrf and before gathering data I wanted to understand how it works...so it was just a short example to describe the issue. Ok, no way to consider real (usage) frequencies by using SPSS Turf...so maby I have to search for another method...guess I am looking for some kind of portfolio optimizer including costs, revenue....and reach. Have a nice weekend! Thanks again. Sven Jon Peck [via SPSSX Discussion] <[hidden email]> schrieb am 18:50 Donnerstag, 26.Januar 2017:

Free forum by Nabble  Edit this page 