The world is linear!
The truth is never linear!
Two new extension commands are now available to help with nonlinearities in regression models.
STATS SPLINES generates three kinds of spline bases for variables. The generated variables can be used in regression and other procedures. It supports bsplines of any order, natural cubic splines, and orthogonal polynomials. Knot placement is automatic. The output is a dataset containing the spline variables.
STATS GAM provides generalized additive models. The terms in the model can be a combination of linear terms, smoothing splines, loess, bsplines, natural splines, or orthogonal polynomials, all generated within the procedure. You can select the error distribution from gaussian, binomial, gamma, and others with various link functions.
Both procedures have custom dialogs, and syntax help is available via F1 in the Syntax Editor or by runn command name /help.
Also recently added was the STATS CORRELATIONS extension command that provides confidence intervals. It requires the Python Essentials normally installed with Statistics.
A small update was made to SPSSINC MODIFY TABLES. In Statistics 24, the significance markers can be included in the main table. The new custom function spreadsig can move these to their own rows. The updated version, 1.6.2, is now on the old site and will be propagated to the new community for V24. Also, with the new markers, MODIFY TABLES can highlight significant cells in various ways.
--
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Locked
