Prediction and Classification with Neural Network Models Export

@article{citeulike:1335259, abstract = {This article compares neural network models to the logit and probit models, the most widely used choice models in current empirical research, and explores the application of neural network models to social science choice/classification problems. Social and political relationships are generally characterized by nonlinearity and complexity and are usually of unknown functional forms. The logit and probit models assume exact and, in general, linear functional forms for the utility functions underlying the observed categorical data. Neural network models, on the other hand, are capable of approximating arbitrary functional forms under general conditions and can handle rich patterns of nonlinearity in the utility functions. They are therefore potentially better suited to typical social science data than the logit and probit models, which are shown to be special cases of the neural network class. Simulation results show that the neural network models perform significantly better than the logit models and are indistinguishable from the "true" models. 10.1177/0049124199027004002}, author = {Zeng, Langche}, citeulike-article-id = {1335259}, citeulike-linkout-0 = {http://dx.doi.org/10.1177/0049124199027004002}, citeulike-linkout-1 = {http://smr.sagepub.com/cgi/content/abstract/27/4/499}, day = {1}, doi = {10.1177/0049124199027004002}, journal = {Sociological Methods Research}, keywords = {classification, forecasting, methodology, neural-net}, month = {May}, number = {4}, pages = {499--524}, posted-at = {2007-05-26 12:55:34}, priority = {5}, title = {Prediction and Classification with Neural Network Models}, url = {http://dx.doi.org/10.1177/0049124199027004002}, volume = {27}, year = {1999} }

TY - JOUR ID - citeulike:1335259 L3 - citeulike-article-id:1335259 N2 - This article compares neural network models to the logit and probit models, the most widely used choice models in current empirical research, and explores the application of neural network models to social science choice/classification problems. Social and political relationships are generally characterized by nonlinearity and complexity and are usually of unknown functional forms. The logit and probit models assume exact and, in general, linear functional forms for the utility functions underlying the observed categorical data. Neural network models, on the other hand, are capable of approximating arbitrary functional forms under general conditions and can handle rich patterns of nonlinearity in the utility functions. They are therefore potentially better suited to typical social science data than the logit and probit models, which are shown to be special cases of the neural network class. Simulation results show that the neural network models perform significantly better than the logit models and are indistinguishable from the "true" models. 10.1177/0049124199027004002 IS - 4 JF - Sociological Methods Research EP - 524 TI - Prediction and Classification with Neural Network Models VL - 27 SP - 499 KW - classification KW - forecasting KW - methodology KW - neural-net AU - Zeng, Langche PY - 1999/05/01/ UR - http://dx.doi.org/10.1177/0049124199027004002 ER -