1 AIT Asian Institute of Technology

A forecasting system based on an object oriented approach

AuthorRegmi, Bhaskar Kumar
Call NumberAIT Thesis no. CS-95-25
Subject(s)Forecasting
Object-oriented programming (Computer science)

NoteA thesis submitted in partial fulfillment of the requirements of the degree of Master of Science
PublisherAsian Institute of Technology
AbstractThis thesis introduces the use of object oriented design technique in the area of forecasting and describes complete software development cycle used in developing the "Object Forecaster. The "Object Forecaster" has the capability of generating three kinds of accuracy indicators for fifteen models simultaneously in a tabular form and offers a convenient way to compare all the forecasting models at a glance, which most of the commercial software lacks. Beside this feature, the system automatically selects optimal smoothing constants and size of moving averages used. Thus forecast values sought from specific models are the most accurate ones. The other normal features are generating forecasts and corresponding residuals, performance indices, descriptive statistics of data used, and graphs for data series, forecasts, residual, autocorrelation function (ACF) and partial autocorrelation function (PACF). Four case studies based on bank data had been performed using "Object Forecaster". This revealed that sophisticated models were not automatically superior and some fairly simple models in certain instances outperformed the sophisticated ones.
Year1995
TypeThesis
SchoolSchool of Engineering and Technology (SET)
DepartmentDepartment of Information and Communications Technologies (DICT)
Academic Program/FoSComputer Science (CS)
Chairperson(s)Huynh, Ngoc Phien;
Examination Committee(s)Nagarur, Nagendra N.;Do, Ba Khang
Scholarship Donor(s)Government of Finland;
DegreeThesis (M.Sc.) - Asian Institute of Technology, 1995


Usage Metrics
View Detail0
Read PDF0
Download PDF0