Gerhard's Extension Nodes for SAS Enterprise Miner

From sasCommunity

Jump to: navigation, search

Image:ExtensionNode_DataPrep_Large.jpg

Contents

Introduction

  • This page contains content of Extension Nodes for SAS Enterprise Miner that are created by Gerhard Svolba.
  • Extension nodes for are a powerful tool to enhance the functionality of SAS Enterprise Miner.
  • At this page you will find two groups of extension nodes
    • Extension Nodes for Data Preparation for Analytics which are based on the book with the same name
    • Various extension nodes that I frequently use in SAS Enterprise Miner (GTOOLS)

Downloads

  • Download the ZIP-File that contains the extension nodes (XML-File, GIF-File, SAS Catalog). Additionally the ZIP file contains
    • two example EM-Diagrams. These diagrams can be imported into SAS Enterprise Miner in order to see an example with the extension nodes.
    • the SAS datasets USAGE and ACCOUNTS
  • Download a paper that describes the use of extension nodes for data preparation for analytics
  • Download a paper that describes the use of the GTOOLS
  • Comments and suggestions can be sent to the author.

Extension Nodes for Data Preparation for Analytics

Image:ExtensionNodes_DataPrep_Small.jpg

  • Based the book Data Preparation for Analytics five node are supplied that allow to pre-process transactional data.
  • Each node outputs the data in a one-row-per-subject data structure so that the data can be used for data mining
  • The following nodes are available
    • Trend Regression: Calculates derived variables that describe the trend of an interval variable in up to two time intervals and creates a concatenated group variable
    • Correlation: Calculates derived variables that describe the correlation of values with its overall mean per timeid or with other input variables
    • Concentration: Calculates derived variables that describe the concentration of an interval variable in a sub-hierachy of the analysis subject
    • Category Count: Calculates derived variables for categorical data. Aggregations like counts, distinct counts or proportions are calculated
    • Transpose from LONG to WIDE: Transpose a dataset from a multiple-row-per-subject structure into a one-row-per-subject structure

Various Extension Nodes - GTOOLS

Image:ExtensionNodes_Gtools_small.jpg

  • These nodes are a collection of SAS Enterprise Miner Tools that I frequently use, as they seem helpful for me:
    • Target BarChart: Displays the relationship between an input variable and the target variables in a bar chart
    • Item-Relationship-Map: Displays the 2-way association rules in a hierarchical tree structure.
    • Anonymous Node: Removes the id variable of a subject from the table and stores it in a separate table. Optionally creates a surrogate key. This node shall be used if a dataset shall be made anonymous.
    • EM DataSource Copy: This node is not yet finished! It will copy to metadata of an existing data source in SAS Enterprise Miner.

Download

  • Read a paper describing Gerhard Svolba's extension nodes
  • Download the files to install Gerhard's extension nodes
  • Download a SAS Enterprise Miner 5.3 demo project to see how the nodes can be used

Links

Personal tools