Re: programatically creating and testing successive models

• To: mathgroup at smc.vnet.net
• Subject: [mg102850] Re: [mg102849] programatically creating and testing successive models
• From: "David Park" <djmpark at comcast.net>
• Date: Sun, 30 Aug 2009 06:07:13 -0400 (EDT)
• References: <13121387.1251543323226.JavaMail.root@n11>

```Just a couple little pieces of help. You can copy and paste the statements
into you Mathematica notebook.

Here is a set of 300 values of test data (with output display suppressed):

data = Table[RandomReal[], {300}];

(Look up Table in Help, also Array.)

Here are the first 100 values:

data[[1 ;; 100]]

Or alternatively:

Part[data, 1 ;; 100]

(Look up Part in Help.)

Here are the values from 100 to 125:

data[[100 ;; 125]]

For data fitting, look up "data fitting" in the Documentation Center SEARCH
box. And maybe "statistics". Maybe others will give you help on that aspect
of the problem.

(If you are truly new to Mathematica it will pay to spend some time learning
the basic syntax and usage of Mathematica before tackling a problem that
might be rather complex.)

David Park
djmpark at comcast.net
http://home.comcast.net/~djmpark/

From: jeff [mailto:jeffreykr at gmail.com]

hi i'm new to mathematica. and have some background in using other
products like minitab and sca for creating models in statisitcs.

i am interested in understanding how one might create models against
different sequential subsets of data. the application is time series
forecasting. first i'd like to create a model to fit data points
1..100, then see its accuracy against points 100..125. then create a
model for points 26..125 and see its accuracy against points 126..150.
and repeat this successively.

how might one do this in mathematica? i have just started looking
through all the vast sections in here and there are lots. so i'm just
asking for direction - which modules to load if any, what programming
constructs to use, which say 3 sections of the documentation would get
me started, any hints about how to get the data into groups that can
be easily worked with for this application.