Combining database management and statistical subroutinesunto a user-oriented data analysis facility (abstract only)

Most database management systems (DBMS) offer convenient and flexible data structures and very good data maintenance facilities. At the same time their data manipulation languages are usually limited and most data analysis applications require extensive programming in a host language. On the other hand, the packages of statistical subroutines (PSS) usually have very good data manipulation and analysis facilities, while at the same time they lack the well known advantages of DBMS. An attempt was made to combine the data definition and maintenance facilities of DBMS and the data manipulation and analysis facilities of PSS into a single user-oriented system. The additional software developed for this purpose performs the following functions:1. Allows the user to define his own analysis and (optionally) store it in a library for further reference.2. Allows the user to define the data on which analysis is to be performed.3. Allows the user to execute the (predefined) analysis in the following way:a. the user's description of analysis is translated into a sequence of data analysis and/or data manipulation subroutines of PSS.b. the required data are retrieved from a database under the control of DBMS and put into a temporary file, whose structure is determined by the analysis input requirements.c. the analysis is performed under the control of the executing program PSS.The outlined system is now being implemented, as an interdepartmental effort, in The Institute for Organization of the Medicine Industry, Warsaw, Poland. It utilizes IMS/VS as a database management system and the OSIRIS III package for statistical data processing. Although the system is still under development it is used not only by research workers but also by administrators and management for relatively simple analysis which are not routinely performed, standard reports. The main advantages of the outlined approach can be summarized as follows:a. reduction of time and cost of preparing the analysis,b. increasing the reliabilityc. presenting the tool for the actual decision makers to perform their own data analysis without the interference with programmers.