Data Structure Examples |
Data structures for data mining data sets and data marts include
Regression Data (see Regression Data), which has the following characteristics:
one observation per customer.
a binary target (dependent) variable, a continuous (interval) target variable, or both. Often only one target variable is modeled per run, but you can use a Group Processing node to model multiple targets. An example of a binary target variable is Purchase or No-purchase, which is useful for modeling customer profiles. An example of a continuous (interval) target variable is Value of Purchase, which is useful for modeling the best customers for particular products, catalogs, or sales campaigns.
many or few input variables, such as ID (identification) variables, demographic data, and history or previous purchases.
cross-sectional data, which is data that is collected across multiple customers, products, geographic regions, but that typically is not collected across multiple time periods.
Association Discovery Data (see Association Discovery Data), also called basket analysis data, which is a common data structure of online transaction processing systems (OLTPs). Such data structures have the following characteristics:
multiple observations per customer. The data set must contain a separate observation for each transaction.
a target (dependent) variable - often a class variable that has a level for each product.
one or more ID variables to identify the customers.
a sequence variable if you are performing a sequence discovery. The sequence variable identifies the ordering of the purchases.
Time Series data analysis, which is more fully supported in SAS/ETS (econometric and time series) software. Analyses that use SAS/ETS procedures can be submitted in the SAS Code node.
Time Series Cross-Sectional (panel) data analysis, which is more fully supported in the SAS/ETS and SAS/STAT products. Analyses that use SAS/ETS and SAS/STAT procedures can be submitted in the SAS Code node.
Copyright © 2006 by SAS Institute Inc., Cary, NC, USA. All rights reserved.