Neural Network Models (Statistical Associates "Blue Book" Series Book 46)

Neural Network Models (Statistical Associates "Blue Book" Series Book 46)

A graduate level introduction to and illustrated tutorial on neural network analysis. <br /> <br />Why we think it is important: Neural network analysis is a valuable tool for prediction of continuous target variables or classification of categorical target variables. It is robust for noisy and missing data, and is particularly useful when nonlinear relationships which cannot be addressed through data transformations or generalized link functions exist in the data. <br /> <br />New title in 2014: <br />* A thorough discussion of implementation of neural network mod4els, including multi-layer perceptron (MLP or backpropagation) and radial basis function (RBF) models. <br />* Illustrates neural network modeling using SPSS and SAS, and explains Stata limitations. <br />* Illustrates use of neural network modeling with SAS Enterprise Miner, which allows automated comparison of fit across various neural and regression models. As such this volume provides an introduction to use of the SAS EM data mining system. <br />* Worked examples with links to data used. <br /> <br /> Below is the unformatted table of contents. <br /> <br /> <br />NEURAL NETWORK MODELS <br />Overview 6 <br />Data examples 8 <br />Artificial neural network software 9 <br />Key concepts and terms 10 <br />Abbreviations 10 <br />Types of artificial neural network models 10 <br />Multilayer perceptron (MLP) models 10 <br />Radial basis function (RBF) models 11 <br />Kohonen self-organizing models 11 <br />Networks, nodes, and weights 13 <br />Models 16 <br />Datasets 16 <br />Training, recall, and learning 17 <br />Training dataset considerations 18 <br />Setting learning parameters 20 <br />Convergence 22 <br />Activation functions 23 <br />Normalization 24 <br />Multilayer perceptron (backpropagation) models 25 <br />Overview 25 <br />MLP models in SPSS 26 <br />SPSS input for ANN-MLP 26 <br />SPSS output for ANN-MLP 40 <br />MLP models in SAS Enterprise Miner 49 <br />Overview 49 <br />Overview of SAS Enterprise Miner steps 50 <br />MLP flow chart 60 <br />Data Partition 60 <br />Modeling 61 <br />Architecture 62 <br />Optimization 63 <br />Model selection criterion 65 <br />Output 66 <br />Model Comparison 73 <br />Scoring 75 <br />MLP Models in SAS PROC NEURAL 77 <br />Overview 77 <br />SAS syntax 77 <br />SAS output 78 <br />Autoneural models in SAS 84 <br />Overview 84 <br />Example 85 <br />Radial basis function models 86 <br />Overview 86 <br />RBF models, data order, and randomization 87 <br />ANN-RBF models in SPSS 88 <br />SPSS input for ANN-RBF 88 <br />SPSS output for ANN-RBF 97 <br />ANN-RBF models in SAS 109 <br />Overview 109 <br />Example using SAS Enterprise Miner 110 <br />Neural network modeling in Stata 112 <br />Assumptions 112 <br />Data level 112 <br />Adequate variance 112 <br />Representative training cases 113 <br />Randomization 113 <br />Few outliers 113 <br />Frequently asked questions 113 <br />What are the “NIST Studies” in relation to ANN? 113 <br />What is a backpropagation model? 114 <br />How can I tell if my results are significant? 116 <br />How can I improve the generalization of my model? 117 <br />Explain neural weights 118 <br />Explain activation (transfer) functions 119 <br />Explain settings for learning rate parameters 121 <br />What are strategies for model complexity vs. model parsimony? 123 <br />Explain quartile analysis 124 <br />Is generalized ANN available? 125 <br />Do I need to transform my input variables? 125 <br />Do I need to standardize my input variables? 125 <br />How should I code binary variables? 127 <br />How do I handle “DK= Don’t Know” and similar codes for my dependent variable? 127 <br />What are pretrained networks? 128 <br />What is a PNN model? 128 <br />What is a GRNN model? 128 <br />What are “constructive algorithms” in ANN-RBF? 129 <br />What software is available to implement ANN models? 129 <br />What are some drawbacks to use of ANN? 129 <br />Bibliography 132 <br />Appendix A: SAS Optimized Data Step Code 136 <br />Appendix B: SAS Results for the “Score” node 141 <br />Pagecount: 144 <br />