Generic model of software cost estimation: A hybrid approach

Software companies ensure to complete the project within time and cost, for which good planning and thinking is required. Software project estimation is a form of problem solving which cannot be solved in a single piece of data by using some formulae. Decomposition of the problem helps in concentrating on smaller parts so that they are not missed. It aids in controlling and approximating the software risks which are commendably fixed and accurate. This paper represents an innovative idea which is the working of Principal Component Analysis (PCA) with Artificial Neural Network (ANN) by keeping the base of Constructive Cost Model II (COCOMO II) model. Feed forward ANN uses delta rule learning method to train the network. Training of ANN is based on PCA and COCOMO II sample dataset repository. PCA is a type of classification method which can filter multiple input values into a few certain values. It also helps in reducing the gap between actual and estimated effort. The test results from this hybrid model are compared with COCOMO II and ANN.