Customer Reviews

Data Mining in Finance: Advances in Relational and Hybrid Methods (The Springer International Series in Engineering and Computer Science)

5 star:		(1)
4 star:		(3)
3 star:		(1)
2 star:		(0)
1 star:		(1)

 

 

I posted the book's table of contents - see "Customer images" link above - and would like to point out Chapter 4, which discusses "relational data mining", also recommending the closely related Chapters 3 and 7. (But not Chapter 2; the authors' mastery of time-series methods - if these do belong to data mining - is not obvious).

I am skeptical about applicability of the proposed method, and would consult other books on neural nets or fuzzy logic - or Bayesian belief networks, for example, forgotten here - but others may disagree.

Jeers to Kluwer for charging over $100 for a book that really should cost $50-60, yet not hiring a proof-reader. ("Daiwa Secretes"!)

DATA MINING IN FINANCE contains a number of practical examples of forecasting S&P 500, exchange rates, stock directions, and rating stocks for portfolio, allowing interested readers to start building their own models. This book is an excellent reference for researchers and professionals in the fields of artificial intelligence, machine learning, data mining, knowledge discovery, and applied mathematics.

It is a very informative book with all major data mining methods and their comparisons compressed into 300 pages. Therefore, a significant part of the book is not leisurely reading. This is typical for the books from Kluwer Academic Publishers. One has to be ready to spend enough time to go through algorithms' details, pseudo code and comparisons of algorithms to get a serious benefit for the design of one's own model.
For instance, understanding the power of first-order if -then rules over the decision trees gained from the book can significantly change and improve design.

An interesting book even if the focus on finance on data mining put the reader always at the border line with some very usual statistics techniques. Literature on ARIMA in finance is exponential for instance.
It will be interesting that the authors develop some examples of the cases on existing and major softwares from the market as Clem or SEM
To read if you need to fulfil some knowledge in finance statistic models more than in data mining in finance, for statistic some Sage papers will give you a more pragmatic over view and for data mining read Larose's books too.

This is one of the most informative books I've found on the subject of mathematical modeling of financial time series. The book is largely a review of the 'state of the art' and frequently expects the reader to be familiar with or willing to 'find and read' relevant articles, but we can all do that, can't we?

The book sequentially studies
1. Standard ARIMA (autoregressive models) which are closest to familiar linear regression techniques.
2. Neural nets and Bayesian trees (as a category called 'relational data mining' by the authors)
3. Fuzzy logic approaches (described as 'membership functions'. Membership functions are defined in terms of linguistic practice, whatever that is.).

In this way, the authors develop a seemingly comprehensive outline of the field, describing fields of study in terms of increasing abstraction. Of the three, I found the fuzzy logic discussion the most interesting.

I have to express some reservations regarding the perspective taken by the authors. Their view is that of the Newtonian physicist observing the interactions of bodies entirely independent of the viewer. At no point do the authors examine the implication of 'self participation' in the marketplace. For example, what happens to probability distribution 'X' when a trading entity uses the probability distribution 'X' to take a significant position in a security? If this seems interesting, you might try looking at "Theory of Financial Risks: From Statistical Physics to Risk Management", by Bouchaud or "An Introduction to Econophysics: Correlations and Complexity in Finance" by Mantegna and Stanley.

This book is badly written. It contains many useless comparisons
between different methods without telling you how to achieve the
best result. You still on your own.