How to group variables for regression analysis?

DarrylR

New Member
Hi all,

Here's my problem: I have 6 variables: Sex (1,2), Age (18, 18, 19, 21, 25, 25, etc), and four variables describing physical parameters; W, X, Y are continuous (1.2323, .0423, etc), and Z is a mean with a standard deviation. FYI, these are bone measurements.

These four variables (W,X,Y,Z) correlate with Sex and/or Age, and I am wanting to see if I can predict age groups, either by sex or combined sex by these four variables.. Is there a way to run an analysis (cluster?) to tell me the best age groupings that give the highest predictive power? I don't want to arbitrarily assigne age groupings, i.e., 0-20, 21-35, 36-50..., because I could test any number of combinations without getting one that works, but don't know if SPSS can do that, or what tests to run. Really new to SPSS, so any thoughts are greatly appreciated!

DarrylR

New Member
As I appreciate the forum and its policies, THIS IS NOT HOMEWORK. I do appreciate the views, but I would appreciate some help even more. I simply have no one with statistical knowledge to ask, And this is a bit more difficult than a simple student t-test or correlation.

md azzainuri

New Member
may be it can help you, if you want run cluster analys on spss
From the Analyze menu, select Descriptive Statistics submenu then Descriptives
Variable (s) box: enter the variable, save and mark the standardized values ​​as variables, and then click OK
then the value of standardization will enter the data into SPSS with the prefix Z. ...
Having obtained the standard values ​​continued to cluster analysis
From the Analyze menu, then select submenu Classify K-Means Cluster ..
Variables box: enter a name that has been standardized
Number of clusters box: enter the number of groups to be formed
Method box: mark iterate and classify
then click OK