Functional Size measurement and COCOMO – A synergistic approach

Accurate estimates are essential to the proper management of software development projects; many software estimates have very wide bounds between the lower and upper estimates of effort. The width of estimates bounds is determined by the degree of uncertainty in the available data. One of these sources of uncertainty is the use of previous projects; especially where those projects do not exhibit a high degree of homogeneity with the project being estimated. The most recent version of the COCOMO model enables the use of IFPUG Function Points. This is accomplished by the use of a set of Unadjusted Function Points UFP to Source Line of Code SLOC conversion tables promulgated by Software Productivity Research. Many authors consider the use of these tables may give unsatisfactory results. The reliability of the use of a single UFP/SLOC conversion figure will be explored, and shown to be subject to a great deal of variation. Additionally the conversion tables are not available for the other functional sizing methods in use today, namely MKII, NEFPUG, COSMIC-FFP. The paper will propose an alternative use of the COCOMO model to assist in the task of estimation. The generally accepted method of estimation using a functional sizing method is to base the estimate on previous project data, where those projects for a homogeneous set with the project under study. The chief difficulty is to find a sufficiently homogeneous set of projects. Research previously carried out can demonstrate that by increasing the degree of homogeneity amongst a set of projects leads to a useful reduction in the variation of the estimates. The proposal is that we may sensibly use the COCOMO cost drivers to allow us to determine a set of homogeneous projects by using a technique derived from estimation by analogy. In addition the COCOMO cost drivers may be used to allow the estimator to adjust his estimates based on the differences between the cost drivers exhibited by the available data and the project under study. This paper is the result of ongoing research and it is offered as a position paper showing the results obtainable under research conditions. The author will be keen to establish links with practitioners to undertake field trials of the proposed approach.