Use this URL to cite or link to this record in EThOS:
Title: Statistical techniques in credit scoring
Author: Whitehead, Christopher David.
ISNI:       0000 0001 3567 1806
Awarding Body: University of Lancaster
Current Institution: Lancaster University
Date of Award: 2005
Availability of Full Text:
Access from EThOS:
The credit industry requires continued development and application of new statistical methodology that can improve aspects of the business. The first part of the thesis suggests a new diagnostic, derived from Kalman filtering, to assess model performance. It allows systematic updating of tracked statistics over time by incorporating new observations with the previous best estimates. It has benefits that current industry practices do not possess and we illustrate its worth on a mortgage application database. The second part of the thesis is concerned with regression analysis of financial data. To aid in the understanding of financial data quantile regression and a variable transformation is applied to a 'missed payments' database resulting in a greater understanding and more accurate description of the data. A less standard sampling and modelling approach is also employed which may give increased predictive power on independent data not used for model construction. The third part of this thesis is concerned with regression modelling in situations where the dimensionality is large. Latent variable modelling of explanatory and binary response variables is suggested which can be maximised using an EM algorithm. Less progress than anticipated has been accomplished in this area. The first two parts of this thesis have suggested novel statistical methodology that can provide benefits over current industry practices, both of which are adapted to real credit scoring applications.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available