|
view
back cover |
|
A
Quick Peek of the Book
·
Table
of Contents
·
Introduction
·
Sample
Chapter |
|
Extra
Chapters of my Book |
|
·
Insight
Node
· Decision
Tree Interactive Training |
|
Contents
of my Book on this Page |
|
|
|
SAS
Data
Mining Links |
|
·
SAS Data
Mining Links |
|
|
|
Highlights
of the Book |
|
· |
An
overview to the sampling nodes in SAS® Enterprise Miner
v4.3 to read the data and learn about the various methods to both
randomly sample and partition the data within the process flow in
SAS® Enterprise Miner v4.3.
|
|
· |
An
overview to the explore nodes that are usually used as a preliminary
step in data mining in order to observe the distribution or range of
values to each variable separately or as a group in discovering
relationships or patterns in the data from the assortment of charts
and graphs that are generated from the nodes. In addition, market
basket analysis can be performed in order to determine the strength
of the association between various items of interest.
|
|
·
|
An
overview to the modify nodes that are designed to perform
preliminary modification to the data set in preparation to data
mining analysis by removing extreme values or outlandish
observations, estimating missing observations in the data,
collapsing and massaging the data in preparation to time series
modeling, performing clustering analysis in determining various
groupings and characteristics within each group that is created in determining the best groupings to form in the data based on a
binary-valued variable.
|
|
·
|
An
overview to the modeling nodes that will make the readers get
familiar with the assortment of modeling designs that can be
performed within Enterprise Miner such as traditional regression
modeling, logistic regression modeling, decision tree modeling,
neural network modeling, principal components modeling, and nearest
neighbor modeling. The book will introduce the readers to an
additive nonlinear modeling design called dmneural network modeling.
The model is designed to predict either an interval or binary target
variable in which the predicted values are calculated from three
separate models. The reason it is called an additive nonlinear
modeling design is because the first model will predict the target
variables, with the subsequent models predicting the residual values
from the previous model. The fitted values are calculated by adding
the separate nonlinear models. In addition, the book will introduce
the readers to a unique two-stage modeling design that can be
performed within SAS® Enterprise Miner. The two-stage
modeling is designed to fit two separate models in succession by
combining the probability estimates from the initial classification
model to predict the continuous response variable in the subsequent
predictive model. In addition, the book will introduce a very
powerful modeling node in Enterprise Miner called ensemble modeling.
Ensemble modeling is designed to increase the accuracy, stability
and consistency of the modeling estimates by either averaging the
prediction estimates from different models or averaging the
prediction estimates based on successive fits of the same predictive
model. In addition, the Ensemble node has the added
capability to perform a couple resampling techniques called bagging
and boosting resampling. Bagging is similar to bootstrap sampling in
which you may iteratively generate your modeling estimates by
resampling the data with replacement, then averaging the fitted
values that results in the final modeling estimates. Boosting is a
classification modeling technique in which the estimated
probabilities are adjusted by weight estimates where the weights are
increased when the classification model misclassified the target
categories from the previous model.
|
|
·
|
An
overview to the assessment nodes that will allow you to evaluate and
assess the results that are generated from the numerous models
within the process flow by viewing the various performance charts
that are generated to determine which classification model performs
best to determining the accuracy of the predictive models from the
various assessment statistics and diagnostic charts that are
generated from the node. The book will introduce the readers to the Reporter
node that assembles and consolidates the numerous option settings
and result listings from the various nodes in the process flow
diagram into a single HTML report that can be viewed by your
favorite Web browser.
|
|
·
|
The remaining sections to the book
will explain the rest of the other nodes in Enterprise Miner such as
the Score node that will allow you to manage, edit, execute
and create your own custom-designed score code in order to generate
entirely different estimates from a new set of values in data mining
analysis. This will be followed by the SAS Code node that is
one of the most powerful nodes within SAS® Enterprise
Miner that will allow you to perform SAS programming within the
process flow.
|
|
|
Programs from Data Mining
Using SAS®
Enterprise Miner |
|
The following is the SAS/IML programming code that is
in regards to my book, Data Mining Using SAS® Enterprise
Miner. |
|
Linear
Regression Modeling
SAS/IML
programming code that computes traditional regression estimates. The SAS/IML®
program will calculate the predicted values, residual values, parameter
estimates and associated standard errors and t-test statistics.
|
|
Logistic
Regression Modeling
SAS/IML
programming code that computes logistic regression estimates. The SAS/IML®
program will calculate the parameter estimates and the likelihood ratio
goodness-of-fit statistics by fitting a binary-valued response variable to
predict based on the maximum likelihood method. An iterative process is
applied in computing the maximum likelihood parameter estimates to
determine the final parameter estimates until convergence.
|
|
K-Means
Clustering
SAS
programming code that computes the k-means clustering estimates. Each
observation is assigned to the cluster with the smallest squared Euclidean
distance based on two separate clusters that are created.
|
|
Principal
Components
SAS/IML
programming code that computes the principal component estimates from the
2004 major league baseball hitters. The SAS/IML® program will calculate the
principal component scores based on the correlation matrix since the
various hitting departments that are measured in different units. A
scatter plot from the first two principal components will be generated in
order for you to observe the variability, outliers and the various
groupings that are formulated from the best hitters in the game of
baseball.
|
|
Articles of Interest |
|
Matignon,
Randall, An Overview of SAS Enterprise Miner
The
paper is designed to make the reader get familiar with the working
environment of SAS® Enterprise Miner v4.3. The paper will
provide you with the general option settings that are available when you
first open SAS® Enterprise Miner v4.3.
|
|
Matignon, Randall, Data
Mining Using SAS Enterprise Miner
The
paper is in reference to my book that is a overview to the multitude of
nodes that are available in SAS® Enterprise Miner v4.3. The
paper will provide you with the purpose of each node, the option settings
that are available and the results that are generated from each node.
|
|
SAS
Institute, Finding the Solution to Data Mining
This
is an update to the subsequent paper on SAS® Enterprise Miner.
.
|
|
SAS
Institute, Finding the Solution to Data Mining
The
paper is in reference to understanding the capability of data mining and
the various nodes that can be used in Enterprise Miner v4.3 to perform
data mining.
|
|
Groth,
Han, and Kamber, SAS Institute, Data Mining
This Microsoft PowerPoint
presentation provides you with a brief overview
to data mining using SAS®
Enterprise Miner v4.3 and the HMEQ home equity loan SAS data set.
|
|
Bommreddy,
Mahesh and Kadiyala, Chaithanya, Data Mining Using SAS Enterprise Miner
The paper is a brief overview
to data mining and the various nodes that are a part of SAS®
Enterprise Miner v4.3.
|
|
E-Commerce
Data Mining Techniques-SAS Enterprise Miner Tutorial
This
short article provides you with a brief description of the working
environment such as the various main menu option
settings in SAS® Enterprise Miner v4.3.
In addition, the paper will introduce you to the various options settings
and and results that are generated from the some of the modeling nodes in SAS® Enterprise Miner v4.3.
|
|
Tom
Bohannon, SAS Institute, Overview of Data Mining
This
course note provides you with a brief overview to data mining and the
SEMMA process using
SAS®
Enterprise Miner v5.
|
|
KPMG
Consulting, Best Practices Approach to the Manufacturing Industry
The paper explains various
SEMMA data mining techniques that are used to address problems in the manufacturing
industry using SAS®
Enterprise Miner v4.3.
|
|
Modeling
Credit Risks: A Practice Lesson for SAS Enterprise Miner
This
course note explains the HMEQ home equity data set that was used in my
book, that is "Data Mining Using SAS Enterprise Miner".
The paper briefly explains the process of constructing a SAS®
project and workspace diagram. The paper compares the classification
performance between logistic regression, neural network and decision tree
modeling using SAS® Enterprise Miner v4.3.
|
|
Bergquist,
David , Evaluation of Significant Variables Between SAS Enterprise
Miner and Tetrad
The
purpose of the paper is to compare the modeling results from various
modeling designs in SAS® Enterprise Miner v4.3 and Tetrad data
mining software in predicting the movement of various toothbrushes.
|
|
Ripley, B.D. , Statistical Data Mining
The
paper is written by B.D. Ripley who is one of the most knowledgeable person in the field of
data mining. The paper explains both traditional clustering and SOM
clustering along with some of the modeling designs that are used in SAS®
Enterprise Miner v4.3.
|
|
SAS
Institute, Data Mining and the Case for Sampling
The
paper explains the various sampling methods that are available in the Sampling
node of SAS® Enterprise Miner v4.3 This paper discusses the use of sampling
as a statistically valid practice for processing large databases. The
paper discusses the advantages and disadvantages of sampling for data
mining, in addition, to explaining the importance of a random sample
in achieving a quality sample.
|
|
Allison,
Paul, Multiple Imputation of Missing Data
The
paper is in reference to the multiple imputation method that is an option
that is available in the Replacement node of SAS®
Enterprise Miner. The Replacement node is designed to impute or
estimate missing values. Multiple imputation uses an appropriate model to
predict the missing values of the variable by all other variables with
non-missing values by iteratively fitting the model numerous times, then
averaging the estimates.
|
|
Bao, Xlinli, Mining
Transaction/Order Data Using SAS Enterprise Miner
The
paper is in regards to the Association node in SAS®
Enterprise Miner. The paper describes the process of analyzing items that
have been purchased from the ASSOCS data set SAS® data set that was used
in my book.
|
|
Neville,
Padraic, Decision Trees for Predictive Modeling
The
paper is in reference to the decision tree modeling that is used in the Tree
node in SAS® Enterprise Miner. The article will explain
the various decision tree modeling methods. In addition, the article will
briefly explain the various ensemble modeling designs such as combining
models, boosting, and bagging resampling.
|
|
SAS
Institute, The ARBORETUM Procedure
The
paper is in reference to the SAS® data mining procedure that
is used to perform decision tree modeling. The importance of this
paper is that some of the option settings that are available in the Tree
node in SAS® Enterprise Miner v4.3 are explained in the article.
|
|
SAS
Institute, DMNeural Procedure
The
paper is in reference to the SAS® dmneural procedure that is
used to perform dmneural network modeling. The importance of this paper is
that some of the option settings that are available in the the Princomp/Dmneural
node are explained in the article.
|
|
Cox,
James, Multidimensional Binary Search
Trees Used for Associative Searching
The
paper is in reference to the RD-tree partitioning technique that is used
in the Memory-Based Reasoning node in SAS® Enterprise
Miner. The partitioning technique is designed to determine the number of
data points to use in calculating the fitted values in nearest neighbor
modeling. The number of data points to combine is determined by the
smoothing constant that must be provided in the predictive or
classification modeling design. The technique performs binary splits to
the data in which the final partitioning of the data results in a
hypercube of data points that are used in calculating the fitted values.
|
|
Breiman, Leo, Arcing Classifiers
The
paper is in reference to boosting resampling that is used in the Ensemble
node. The paper provides the reader with the formula that is used in
SAS® Enterprise Miner and boosting resampling in which the
weight estimates are calculated that are used to adjust the estimated
probabilities from the classification model to generate the probability
estimates to the boosting model.
|
|
Cerrito,
Patrica B., Comparison of
Enterprise Miner and SAS/Stat for Data Mining
The
paper compares some of the procedure output listings from various
statistical procedures that are available in SAS® for data
mining with the results that are generated from some of the SAS®
Enterprise Miner nodes.
|
|
Gallaugher, John, Modeling
Credit Risks: A Introduction to Data Mining using SAS
Enterprise
Miner
The
paper constructs a process flow diagram for credit risk modeling. The
paper explains the process in constructing separate classification models
such as logistic regression, neural network, and decision tree models,
then assessing the accuracy between the separate models.
|
|
Sarle,
Warren, SAS
Macro Programs for Jacknifing and Bootstrapping
This
link will provide you with various SAS macro programs for jackknifing
and bootstrapping parameter estimation in computing approximate standard
errors, bias-corrected estimates, and confidence intervals, that is assuming
that there is simple
random sampling that is performed.
|
|
Data
Mining Using SAS Enterprise Miner Blog
|
Data
Mining Using SAS Enterprise Miner Blog
The following link is to my personal blog. The purpose of my blog is to
give you an opportunity to respond to me by simply posting any questions that you
might have with my book, that is "Data Mining Using SAS Enterprise
Miner". Otherwise, please feel free to ask any questions that you might
have with SAS® Enterprise Miner. |
|
|
|
|
|
|
|
Ordering
Information
|
|
John Wiley & Sons, Inc.
111 River Street
Hoboken, NJ 07030-5774
http://www.wiley.com/WileyCDA/WileyTitle/productCd-0470149019.html
|
|
|
|
|
|
Back to Home Page |
