COMPARATIVE ANALYSIS OF THE DIFFERENT METHODS OF APPROXIMATION OF ENVELOPE CURVES OF THE REFERENCE POINTS FOR THE DEMAND FOR VARIOUS INFORMATION RESOURCES IN AN ACADEMIC LIBRARY

The present research describes a possible solution to the problem of demand forecasting for a specific type of queuing. The current researchers investigate academic library patrons’ demand for information resources pertaining to specific subject headings. This exploratory study is conducted in the Charles Evans Inniss Memorial Library of the Medgar Evars College of the City University of New York.

The problem of prediction of a random process is not new in the field of Information Science. This problem is usually considered from the perspective of analytical methods based on results of statistical modeling. Another plausible approach entails self-learning systems based on neural networks. However, real-life experience suggests that in many libraries the demand prediction and collection development do not always correlate. Implementation of such a method would minimize the number of failures to meet customer demand, especially during peak periods.

Collection development in an intuitive fashion might result either in the budgetary over-expenditure or in increase of waiting time in case of insufficient number of copies of the document. The latter problem is often associated with the possible increase in delays in library services and general deterioration of the service quality. In academic libraries delays of service lead to a significant slowdown in scholarly and research activity. This kind of problem could result in a possibility of not meeting deadlines for grants and stipends by the researchers.

This research reports on a method, developed by the authors, which is designed to address the prediction of demand for specific subject headings and some of their segments. The efficiency and effectiveness of this forecast of demand can be achieved by means of retrospective approximation of demand curves, utilizing specific functions, both analytical and computer-simulated. Furthermore, a comparative analysis of the results of approximations will be conducted, as well as analyses of correlations between the predicted demand values and the actual values.