Catalogue Search | MBRL
Are you sure you want to remove the book from the shelf?
{{itemTitle}}
17
result(s) for
"Coggeshall, Stephen"
Sort by:
Foundations of predictive analytics
\"Preface this text is a summary of techniques of data analysis and modeling that the authors have encountered and used in our two-decades experience of practicing the art of applied data mining across many different fields. The authors have worked in this field together and separately in many large and small companies, including the Los Alamos National Laboratory, Bank One (JPMorgan Chase), Morgan Stanley, and the startups of the Center for Adaptive Systems Applications (CASA), the Los Alamos Computational Group and ID Analytics. We have applied these techniques to traditional and nontraditional problems in a wide range of areas including consumer behavior modeling (credit, fraud, marketing), consumer products, stock forecasting, fund analysis, asset allocation, and equity and xed income options pricing. This monograph provides the necessary information for understanding the common techniques for exploratory data analysis and modeling. It also explains the details of the algorithms behind these techniques, including underlying assumptions and mathematical formulations. It is the authors' opinion that in order to apply di erent techniques to di erent problems appropriately, it is essential to understand the assumptions and theory behind each technique. It is recognized that this work is far from a complete treatise on the subject. Many excellent additional texts exist on the popular subjects and it was not a goal for this present text to be a complete compilation. Rather this text contains various discussions on many practical subjects that are frequently missing from other texts, as well as details on some subjects that are not often or easily found. Thus this text makes an excellent supplemental and referential resource for the practitioners of these subjects\"--Provided by publisher.
Book
Foundations of Predictive Analytics
Drawing on the authors' two decades of experience in applied modeling and data mining, this self-contained book presents the fundamental background required for analyzing data and building models for many practical applications, such as consumer behavior modeling, risk and marketing analytics, and other areas. It explains the algorithmic details behind each technique, including underlying assumptions and mathematical formulations, and discusses a variety of practical topics that are frequently missing from similar texts. Software and examples are available at www.DataMinerXL.com.
eBook
Properties of Statistical Distributions
This chapter presents common distributions widely used in data analysis and
modeling. Fully understanding these distributions is key to understanding the
underlying assumptions about the distributions of data. The next chapter
will discuss various aspects of matrix theory, which is a convenient vehicle to
formulate many of these types of problems.
Book Chapter
Time Series Analysis
Modeling for time series is conceptually similar to other modeling problems, but one major distinction is that usually the next value of the series is
highly related to the most recent values, with a time-decaying importance in
this relationship to previous values. Because of this property a different set of
machinery has evolved for the special case of time series modeling, sometimes
loosely called forecasting or even prediction.
Book Chapter
Linear Modeling and Regression
The most common data modeling methods are regressions, both linear and
logistic. It is likely that 90% or more of real world applications of data
mining end up with a relatively simple regression as the final model, typically after very careful data preparation, encoding, and creation of variables.
There are many kinds of regression: both linear, logistic and nonlinear, each
with strengths and weaknesses. Many regressions are purely linear, some only
slightly nonlinear, and others completely nonlinear. Most multivariate regressions consider each independent variable separately and do not allow for nonlinear interaction among independent variables. Treatment of nonlinearities
and interactions can be done through careful encoding of independent variables such as binning or univariate or multivariate mapping to nonlinear functions. Once this mapping has been done one can then do a linear regression
using these new functions as independent variables.
Book Chapter
Miscellaneous Topics
In this last chapter we discuss some particular relevant topics in other areas
not covered in the previous chapters.Visualization of high-dimensional data is an interesting, useful yet challenging task. The approach of dimensionality reduction strives to find a lower
dimensional space that captures all or most of the information that exisits in
the higher dimensional space. Multidimensional scaling is a process whereby
we seek an essentially information-preserving subspace of the original space.
That is, we look for a lower dimensional space in which the problem becomes
simpler or more clear, and still retains its essence. This discipline originated as
a way to visualize higher-dimensional data, “projected” into lower dimensions,
typically two, in such a way as the distances between all points was mostly preserved. The process has evolved into general dimensionality reduction methodology as well as being able to discover embedded, lower-dimensional nonlinear
structures in which most of the data lies. Here we explore the general problem of finding a lower-dimensional space in which the data can be represented
and the between-point distances is preserved as best as possible. For a more
complete treatment on multidimensional scaling, refer to Cox and Cox (2000).
Book Chapter
Nonlinear Modeling
In Chapter 4, we discussed many of the important and popular linear modeling techniques. This chapter is devoted to the subject of nonlinear modeling
techniques. We will cover neural networks, decision trees, additive models,
support vector machine (SVM), and fuzzy logic systems.
Book Chapter
Data Preparation and Variable Selection
An important philosophy in machine learning is that one should not expect
the model to do all the difficult work, not even a powerful nonlinear model.
In practical situations with finite and noisy data, it is useful to encode the
inputs as best as possible using expert knowlege and good statistical practices.
Yes, this generally reduces potential information to the model, but in practice it allows the model to focus its efforts in the right areas. Good thought
to encoding variables is fundamental to successful modeling. Do not expect
your model to do the heavy lifting. Help the model as much as possible with
thoughtful data encoding and expert variable creation.
Book Chapter