Computerized life cycle advising, monitoring, and predicting (CLAMP)

Abstract

Managers of software projects require automated tools for integrating and smoothly carrying out control, advising, monitoring, and prediction management tasks. However, contemporary project managers still can only employ manual tools to measure a programmer's status. These measurements are sometimes inaccurate, causing, directly or indirectly, delay of many software projects. Compounding the problem of inadequate automation is the feeling among managers that programmers are against being measured. The system proposed here, called CLAMP, is designed to computerize the manager's task of advising and monitoring software life cycle. CLAMP tracks the programmer's status by automatically collecting and analyzing a programmer's activities and programs. It also provides technical advice to programmers and project managers. If a programmer history database exists, CLAMP can provide a prediction of project completion status to managers. CLAMP consists of two main subsystems, a data collection subsystem and a Metric-Oriented Life cycle Director (MOLD). The automated data collection subsystem collects quantitative data (metrics) from two sources: source programs and programmer activities. MOLD is an experiment in rule-based system development to interpret metrics collected in the data collection subsystem. CLAMP was evaluated using data collected from two graduate level software projects. These evaluations showed: (1) the chosen metrics were adequate for management purposes in these projects, (2) the automated collection of the metrics was feasible, (3) the expert system (MOLD) can be used for computerizing the software life cycle advising, monitoring, and predicting, (4) the usefulness of rules used to provide technical advice to programmers and project managers, and monitor programmer's status, and (5) the effectiveness of Unix tools in constructing the prototype.