Page 249 -
P. 249

5.14 Neural Networks in Data Mining   237


                                Summarizing some conclusions of these experiments:
                              - Quick searches may not yield the solution that performs best (e.g. GIR  solution
                                for  all  values  of  BRANCH).  This  obvious  conclusion  can  be  dramatically
                                important in data mining applications where "quick search" is a must.
                              - Quick  searches  may  fail  to  detect  the  crucial  importance  of  a  variable
                                (BRANCH in the example).
                              - Solutions that are of interest may require large time-consuming searches. In the
                                example, the  (CA, DEPR, AIC}  solution  could  be  more  interesting  from the
                                economical point of view than the solution using GIR.
                              - Interpretation of the results depends drastically  on the solution. In the example,
                                an obvious relationship exists between CapR and GIR  because both  are ratios
                                with the same numerator, the net income, which is the causal element. A causal
                                inference  for the  {CA, DEPR,  NC) solution  is  more problematic,  to say the
                                least.

                                These  experiments  reflect  the  difficulties  one  may  face  in  real  data  mining
                              applications, especially concerning the usefulness of the mining results.


                               Bibliography


                               Anthony  M,  Bartlett  PL  (1999)  Neural  Network  Learning:  Theoretical  Foundations.
                                 Cambridge University Press.
                               Atiya AF, El-Shoura SM, Shaheen SI, El-Sherif MS (1999) A Comparison Between Neural-
                                 Network Forecasting Techniques. Case Study: River Flow Forecasting. IEEE Tr Neural
                                 Networks, 10:402-409.
                               Baum  EB,  Haussler  D  (1989)  What  Size  Net  Gives  Valid  Generalization?  Neural
                                 Computation, 1: 15 1-160.
                               Berson A, Smith SJ (1997) Data Warehousing, Data Mining and OLAP. McGraw Hill Co.
                                 Inc.
                               Bishop CM (1995) Neural Networks for Pattern Recognition. Clarendon Press, Oxford.
                               Blumer A, Ehrenfeucht A, Haussler D, Warmuth MK  (1989) Learnability and the Vapnik-
                                 Chernovenkis Dimension. J Ass Comp Machinery, 36:929-965.
                               Carter MA, Oxley ME  (1999) Evaluating the Vapnik-Chervonenkis Dimension of Artificial
                                 Neural Networks Using the Poincare Polynomial. Neural Networks, 12:403-408.
                               Cherkassky V, Mulier F (1998) Learning from Data, John Wiley & Sons, Inc.
                               Chryssolouris G,  Lee  M,  Rarnsey  A  (1996) Confidence  Interval  Prediction  for  Neural
                                 Network Models, IEEE Tr Neural Networks, 7:229-232.
                               Cortes C, Vapnik V (1995) Support Vector Networks. Machine Learning, 20:273-297.
                               Cover TM (1965) Geometrical and Statistical Properties of  Systems of  Linear Inequalities
                                 with Application in Pattern Recognition. IEEE Tr Elect Comp, 14:326-334.
                               Eberhart RC, Dobbins RW  (1990) Background and History. In: Eberhart RC, Dobbins RW
                                 (eds) Neural Network PC Tools. A Practical Guide. Academic Press, Inc., pp 9-34.
                               Eberhart  RC,  Dobbins RW  (1990) Implementations. In:  Eberhart RC, Dobbins RW  (eds)
                                 Neural Network PC Tools. A Practical Guide. Academic Press, Inc., pp 35-58.
   244   245   246   247   248   249   250   251   252   253   254