Effect of data standardization on neural network training

Shanker, Murali; Hu, Michael Y.; Hung, Ming S.

doi:10.1016/0305-0483(96)00010-2

Cited by 240 publications

(123 citation statements)

References 5 publications

Supporting

Mentioning

122

Contrasting

Unclassified

Order By: Relevance

“…This so-called data preprocessing is claimed by some authors to be bene®cial for the training of the network. Based on our experience (Shanker et al [56] and also a preliminary study for this project), data transformation is not very helpful for the classi®cation task. Raw data are hence used without any data manipulation.…”

Section: Design Of Neural Network Modelmentioning

confidence: 99%

Artificial neural networks in bankruptcy prediction: General framework and cross-validation analysis

Zhang

Patuwo

et al. 1999

European Journal of Operational Research

539

280

View full text Add to dashboard Cite

In this paper, we present a general framework for understanding the role of arti®cial neural networks (ANNs) in bankruptcy prediction. We give a comprehensive review of neural network applications in this area and illustrate the link between neural networks and traditional Bayesian classi®cation theory. The method of cross-validation is used to examine the between-sample variation of neural networks for bankruptcy prediction. Based on a matched sample of 220 ®rms, our ®ndings indicate that neural networks are signi®cantly better than logistic regression models in prediction as well as classi®cation rate estimation. In addition, neural networks are robust to sampling variations in overall classi®cation performance. Ó

show abstract

Section: Design Of Neural Network Modelmentioning

confidence: 99%

Artificial neural networks in bankruptcy prediction: General framework and cross-validation analysis

Zhang

Patuwo

et al. 1999

European Journal of Operational Research

539

280

View full text Add to dashboard Cite

show abstract

“…The scale is changed by upsampling and downsampling operations. The DWT has many advantages in compressing a wide range of ranges to obtain better training (Shanker et al 1996;Gunn 1998;Singh et al 2009). The appropriate data normalization range for daily flow forecasting using ANN is reported in Three models were proposed for 1-, 2-and 3-day flood forecasting and are presented in Table 2.…”

Section: Discrete Wavelet Transformmentioning

confidence: 99%

Daily river flow forecasting using wavelet ANN hybrid models

Pramanik

Panda

Singh

2010

Journal of Hydroinformatics

View full text Add to dashboard Cite

“…Shanker et al (1996) and Luk et al (2000) reported that networks trained on transformed data achieve better performance and faster convergence in general, although the advantages diminish as network and sample size become large. In this study, the data were transformed with the log function, and the deterministic values of the input parameters were removed prior to modeling.…”

Section: Study Area and Data Setmentioning

confidence: 97%