Neural networks
Article Abstract:
Neural networks can replace the traditional statistical techniques used in score model development for the credit industry. Traditional statistical methods, such as multiple linear and logistic regresssions, require knowledge, training and experience to be effectively used. They are also time-consuming manual processes that slow down model development and its application to automated systems. Neural networks are not hampered by the limitations of traditional statistical techniques and enable complicated, non-linear modeling difficult with conventional methods. Neural networks train themselves to process information by integrating weighted values of inputs, processing elements and outputs to generate a predicted result or score. Network models are included in operational processing systems and allow users to determine unknown factors such as customer behavior.
Publication Name: Credit World
Subject: Banking, finance and accounting industries
ISSN: 0011-1074
Year: 1993
User Contributions:
Comment about this article or add new information about this topic:
The daddy connection
Article Abstract:
Research is debunking the belief that mothers are responsible for birth defects. Research evidence shows that drugs, alcohol and other toxic substances may have infect the embryo through contaminated sperm.
Publication Name: Reader's Digest (Canadian)
Subject: General interest
ISSN: 0034-0413
Year: 1992
User Contributions:
Comment about this article or add new information about this topic:
The write guy for the job
Article Abstract:
Phillips testifies about handwriting in court cases and charges $3,500 to appear as an expert witness. His methods and some cases he has been involved with are discussed.
Publication Name: Philadelphia Magazine
Subject: Regional focus/area studies
ISSN: 0031-7233
Year: 1993
User Contributions:
Comment about this article or add new information about this topic: