Application of Neural Networks to House Pricing and Bond Rating

H.A.M. Daniëls, B. Kamp, W.J.H. Verkooijen

Research output: Working paperDiscussion paperOther research output

424 Downloads (Pure)

Abstract

Feed forward neural networks receive a growing attention as a data modelling tool in economic classification problems. It is well-known that controlling the design of a neural network can be cumbersome. Inaccuracies may lead to a manifold of problems in the application such as higher errors due to local optima, overfitting and ill-conditioning of the network, especially when the number of observations is small. In this paper we provide a method to overcome these difficulties by regulating the flexibility of the network and by rendering measures for validating the final network. In particular a method is proposed to equilibrate the number of hidden neurons and the value of the weight decay parameter based on 5 and 10-fold cross-validation. In the validation process the performance of the neural network is compared with a linear model with the same input variables. The degree of monotonicity with respect to each explanatory variable is calculated by numerical differentiation. The outcomes of this analysis is compared to what is expected from economic theory. Furthermore we propose a scheme for the application of monotonic neural networks to problems where monotonicity with respect to the explanatory variables is known a priori. The methods are illustrated in two case studies: predicting the price of housing in Boston metropolitan area and the classification of bond ratings.
Original languageEnglish
Place of PublicationTilburg
PublisherOperations research
Number of pages18
Volume1997-96
Publication statusPublished - 1997

Publication series

NameCentER Discussion Paper
Volume1997-96

Fingerprint

Neural networks
Costs
Economics
Feedforward neural networks
Neurons
Data structures

Keywords

  • Classification
  • error estimation
  • monotonicity
  • finance
  • neural-network models

Cite this

Daniëls, H. A. M., Kamp, B., & Verkooijen, W. J. H. (1997). Application of Neural Networks to House Pricing and Bond Rating. (CentER Discussion Paper; Vol. 1997-96). Tilburg: Operations research.
Daniëls, H.A.M. ; Kamp, B. ; Verkooijen, W.J.H. / Application of Neural Networks to House Pricing and Bond Rating. Tilburg : Operations research, 1997. (CentER Discussion Paper).
@techreport{a84dd378e1804f07a2dce93524ed7bee,
title = "Application of Neural Networks to House Pricing and Bond Rating",
abstract = "Feed forward neural networks receive a growing attention as a data modelling tool in economic classification problems. It is well-known that controlling the design of a neural network can be cumbersome. Inaccuracies may lead to a manifold of problems in the application such as higher errors due to local optima, overfitting and ill-conditioning of the network, especially when the number of observations is small. In this paper we provide a method to overcome these difficulties by regulating the flexibility of the network and by rendering measures for validating the final network. In particular a method is proposed to equilibrate the number of hidden neurons and the value of the weight decay parameter based on 5 and 10-fold cross-validation. In the validation process the performance of the neural network is compared with a linear model with the same input variables. The degree of monotonicity with respect to each explanatory variable is calculated by numerical differentiation. The outcomes of this analysis is compared to what is expected from economic theory. Furthermore we propose a scheme for the application of monotonic neural networks to problems where monotonicity with respect to the explanatory variables is known a priori. The methods are illustrated in two case studies: predicting the price of housing in Boston metropolitan area and the classification of bond ratings.",
keywords = "Classification, error estimation, monotonicity, finance, neural-network models",
author = "H.A.M. Dani{\"e}ls and B. Kamp and W.J.H. Verkooijen",
note = "Pagination: 18",
year = "1997",
language = "English",
volume = "1997-96",
series = "CentER Discussion Paper",
publisher = "Operations research",
type = "WorkingPaper",
institution = "Operations research",

}

Daniëls, HAM, Kamp, B & Verkooijen, WJH 1997 'Application of Neural Networks to House Pricing and Bond Rating' CentER Discussion Paper, vol. 1997-96, Operations research, Tilburg.

Application of Neural Networks to House Pricing and Bond Rating. / Daniëls, H.A.M.; Kamp, B.; Verkooijen, W.J.H.

Tilburg : Operations research, 1997. (CentER Discussion Paper; Vol. 1997-96).

Research output: Working paperDiscussion paperOther research output

TY - UNPB

T1 - Application of Neural Networks to House Pricing and Bond Rating

AU - Daniëls, H.A.M.

AU - Kamp, B.

AU - Verkooijen, W.J.H.

N1 - Pagination: 18

PY - 1997

Y1 - 1997

N2 - Feed forward neural networks receive a growing attention as a data modelling tool in economic classification problems. It is well-known that controlling the design of a neural network can be cumbersome. Inaccuracies may lead to a manifold of problems in the application such as higher errors due to local optima, overfitting and ill-conditioning of the network, especially when the number of observations is small. In this paper we provide a method to overcome these difficulties by regulating the flexibility of the network and by rendering measures for validating the final network. In particular a method is proposed to equilibrate the number of hidden neurons and the value of the weight decay parameter based on 5 and 10-fold cross-validation. In the validation process the performance of the neural network is compared with a linear model with the same input variables. The degree of monotonicity with respect to each explanatory variable is calculated by numerical differentiation. The outcomes of this analysis is compared to what is expected from economic theory. Furthermore we propose a scheme for the application of monotonic neural networks to problems where monotonicity with respect to the explanatory variables is known a priori. The methods are illustrated in two case studies: predicting the price of housing in Boston metropolitan area and the classification of bond ratings.

AB - Feed forward neural networks receive a growing attention as a data modelling tool in economic classification problems. It is well-known that controlling the design of a neural network can be cumbersome. Inaccuracies may lead to a manifold of problems in the application such as higher errors due to local optima, overfitting and ill-conditioning of the network, especially when the number of observations is small. In this paper we provide a method to overcome these difficulties by regulating the flexibility of the network and by rendering measures for validating the final network. In particular a method is proposed to equilibrate the number of hidden neurons and the value of the weight decay parameter based on 5 and 10-fold cross-validation. In the validation process the performance of the neural network is compared with a linear model with the same input variables. The degree of monotonicity with respect to each explanatory variable is calculated by numerical differentiation. The outcomes of this analysis is compared to what is expected from economic theory. Furthermore we propose a scheme for the application of monotonic neural networks to problems where monotonicity with respect to the explanatory variables is known a priori. The methods are illustrated in two case studies: predicting the price of housing in Boston metropolitan area and the classification of bond ratings.

KW - Classification

KW - error estimation

KW - monotonicity

KW - finance

KW - neural-network models

M3 - Discussion paper

VL - 1997-96

T3 - CentER Discussion Paper

BT - Application of Neural Networks to House Pricing and Bond Rating

PB - Operations research

CY - Tilburg

ER -

Daniëls HAM, Kamp B, Verkooijen WJH. Application of Neural Networks to House Pricing and Bond Rating. Tilburg: Operations research. 1997. (CentER Discussion Paper).