Doctor of philosophy

Principal Components

The principal components procedure was originally introduced by Hotelling (1933) and was only later adopted to regression analysis, Principal components estimates were shown to be superior to OLS estimates in several stimulations. These stimulations however could not guide the analyst when to prefer principal components over OLS in a real setting. A preferred method is to derive the difference between the SMSE of OLS and principal components. If it is positive, use principal principle components.

Utilization of the principle components procedure should be accompanied by careful consideration of its limitations. First number of principles components utilized in the model is reduced by the number of zero or near zero, eigenvalues of f the X’X matrix.

Secondly it is applicable to multiplicative models only if all the observations have positive values fit to a logarithm transformation, Mahajan et al (1977). In addition researchers using principal components should always examine whether they gain any improvement over OLS in terms of SMSE.

The Bayesian Approach

Multicollinearity results in some loss of information concerning the model’s parameter vector. In the Bayesian approach β is a random variable with a known prior distribution. The specification of the prior is tantamount to providing additional information concerning β. The use of Bayesian approach in regression was discussed in Leamer (1973, 1978) and zellner (1971). One of the main criticisms of this approach is concerning its subjectiveness as it requires an exact specification of the prior distribution. In this respect a study of the sensitivity of the posterior mean to changes in the prior distribution, particularly those in the prior variance-covariance matrix, would be very useful. Chamberlain and Leamer (1976) considered a given location of the prior and developed a correspondence between classes of prior variance-covariance matrices and bounded regions in the space of the posterior mean vector was generalized by Leamer (1982) by assuming that the prior variance-covariance matrix is considered to lie between a minimum variance-covariance matrix and a maximum variance-covariance matrix.

A Bayes-like technique, called mixed estimation, was proposed by Theil and Goldberger (1961) and Theil(1963) in which certain prior restrictions are made on β. These restrictions are formulated in terms of an additional model involving β and a random vector independent of the error vector ε. The prior and sample information (provided by the original model) are combined via the method of weighted least squares to produce an unbiased mixed-estimation estimator. The advantage of the mixed estimation technique is that it allows the user to incorporate prior knowledge about β in the model without having to specify a particular prior distribution for β. If, however, normality is assumed and the prior knowledge about β can be used to determine the mean and variance, then a normal prior would be fully specified.

LIST OF PUBLICATIONS & PRESENTATIONS

1. Published on article in the Journal of Institution of Engineers on Computers in Chemical Engineering.

2. Presented a paper in "International Seminar" held at IFMR, in the area of Rural Finance.

3. Monograph on "Neural Network" nearing completion.

4. Presented a paper on “Improving prediction accuracy of loan default- A case in rural credit”, at the national conference of Operations Research Society.

5. The abstract was published in the Operations Research Society proceedings since, the article did not reach them.

6. Was a Co-Chairman in a national level seminar on “Paradigm Shift Of Finance Function”, organized by a business school in Coimbatore.

7. Published a paper titled “Default prediction in Financial services –A Quantitative approach”, In SRM Digest , a journal published by SRM University.

8. A paper titled "An Integrated Model for Credit Risk Assessment In

the Rural Context in India” has been published in Academy of Taiwan Business Management Review, Vol .4,Num.3,December

2008. The ISSN number issued by the ISSN INTERNATIONAL CENTRE, FRANCE is 1813-0534

9. Have made various Presentations to different customers for customer profiling, Positioning, Business Process mapping, Business Solution, Business Strategies etc.

10. Story in Science fiction in progress

SPECIFIC CONTRIBUTION

The improvement in prediction accuracy achieved in this research enables accurate scoring and profiling of customers in financial service

and its importance is globally important as has been proved by the global crisis triggered by Subprime lending.

In the financial services sector and specially in the retail lending sector, as expressed by people in the industry, there is no comprehensive model to profile borrowers which takes all micro, macro economic factors, diversities in the socio, economic, cultural and demographic, psychographic aspects of customers in different regions and in this work an attempt has been made towards this. An integrated model has been developed which is a holistic approach towards behavior of customer. This was reiterated by the appreciation of many top people from the industry during the paper presentation in a seminar and expression of interest in using my services to develop models for their business.

This work could be further used for response modeling since as a method this has been proved in this work, to be very accurate in predictive modeling. This would be a step forward in scoring customers not only for default prediction but to improve ROI in campaigns for upselling and cross selling.

BIOGRAPHY OF A. SUNDARAMURTHY

A. Sundaramurthy started as a Management Trainee in a public sector Engineering Company 28 years ago and has contributed extensively in R&D. engineering, design, simulation and modeling, investment analysis, projects QMS, marketing, HR, strategic management, in various Indian and MNC's both in India and abroad. Some of the companies associated include ICI, Technip, Shell, American Express Bank, Parry, Alpha Laval, General Foods, BG,BP,MODEC International, CST Vulcan, Kelllog, etc.

Rose as Department head, Divisional Head and headed Corporate functions of Technology and HR and also as Country business head of application software division in an American MNC.

Has developed products and processes and have developed implemented various business policies and strategies. Now a Director and Principal Consultant of a Pvt. Ltd. Management Consulting Company offering Technical and Management consulting for businesses in Middle East, US, Australia and India for Financial Sector, FMCG, Telecom, Oil & Gas etc. Some of the Clients include Shell, American Express Bank Financial Services, , Parry ware, Valued Epistemic, a publishing house, an International Energy Co etc.

Is a Director in a Singapore based company in the business of PLM, Business process solutions and engineering services.

Is a Director in Company in the business of data mining, analytics and CRM in collaboration with a Market Research company in Wisconsin.

Have worked in Malaysia, Singapore, Japan, Middle East, USA, UK and European Countries.

Was a faculty for a BITS, Pilani, Distance Learning Program.

Have Co-Chaired a National level Seminar “Financial Services Business” and was a panel member in a discussion titled “Paradigm shift of financial function” where other panel members were GM(FIN) of LMW, Director of Sankara Business School, Director of Chennai Business School and CNBC ,Cochin Correspondent.

SHORT BIO-DATA OF PROF. MJ XAVIER

Prof. MJ Xavier obtained his Doctorate in Management (1984) from the Indian Institute of Management, Calcutta. He is basically an engineer with an M.Tech. (1979) in Chemical Plant Engineering from Regional Engineering College, Warangal and B.Tech. (1976) in Chemical Engineering from Coimbatore Institute of Technology.

He has more than 20 years of professional experience in teaching, academic administration and industry. He has served as a Research Executive in Mode Research Pvt. Ltd., Calcutta (1982-84) and as Manager in-charge of Management Development and Services with SPIC Ltd., Chennai (1985-91). He has taught at XLRI, Jamshedpur (1984-84), IIM Bangalore (1991-96) and IFMR Chennai (1998 till date). For a brief period he served as the Director of the Karunya Management School and the Principal of the Karunya Institute of Technology at Coimbatore. He also served as the Dean of the Academy for Management Excellence (a division of IFMR) for a period of three years (200-03).

His areas of interest include CRM, Marketing Research, Strategic Marketing and Data Mining. He has conducted a number of training programmes for executives and has also been a consultant to several companies.

He has authored three books and published more than 70 articles in Journals and Magazines in India and Abroad. His book `Marketing in the New Millennium’ won the DMA-Escorts Award for the best Management Book of the Year 1999.

He has served as Visiting Faculty to a number of business schools in India and abroad including The University of Buckingham, U.K. (1993), The Texas Christian University, U.S.A. (1999), California Polytechnic State University, USA (2003-05) and The American University of Armenia, Armenia (2006).

Contact Address:
Dr. M.J. Xavier

K.B. Chandrasekar Distinguished Professor of Marketing &

Director – Kotler-Srinivasan Center for Research in Marketing

Great Lakes Institute of Management,

24, South Mada Street , Srinagar Colony,

Saidapet, Chennai-600015.

Telephone: [phone] – Ext:214

Website : www.greatlakes.edu.in

Email: [e-mail] & [e-mail]

1 H.J. Kappen, Foundation for neural Network, Postbus 9101, 6500 HB Nijmegen, The Netherlands. Tel. [phone], Fax. +31 24~3541435