Analysis of Combining Software Estimation Techniques

In this paper, analysis of the result obtained from experiment with three projects is described. Combination of different estimation techniques helps estimator to reduce error and keep control over the deviation of estimates away from actual. As a response to survey from estimators working in software industry, It is revealed that, according to stages, estimates are termed as budgetary, initial, progressive and closure. As a budgetary estimates are dominated by analogy based estimation techniques, It can be complemented by COCOMO II Application Composition Model. The Initial estimates is average of COCOMO II Early Design model with Object point sizing, which can be complemented by Function point and Usecase point based estimation. The Progressive estimates are calculated by averaging of COCOMO II Post Architecture Model and Class point based estimates. It is observed that effort estimates are more accurate than using only COCOMO II.