Software Cost Estimation Based on the Hybrid Model of Input Selection Procedure and Artificial Neural Network

Software effort estimation is the forecasting of development effort and development time needed to develop any software project. It is considered to be the very primary step of software development process and at the same time considered to be the key task as accurate assessments of growth of the current project, its delivery exactness and its cost control can only be achieved once desired estimation is accurate. And at broader perspective an accurate estimation of a currently developing software product will result in landing the organization in a better schedule of its futuristic software projects too. With due above reason, software effort estimation has received a considerable amount of attention of many researchers from past so many decades. In this paper, software cost estimation is done by first performing a proposed input selection procedure to get the relevant set of cost drivers and leaving behind the irrelevant attributes. In the next step, it is now only these relevant set of attributes that are being assigned to Artificial Neural Network as its input for the purpose of getting the accurate estimation of software development effort and cost. Removing the irrelevant cost drivers at the very first step directly leads to attain accurate software cost estimation results. Besides this the proposed model results in a significant decrease of complexities associated with traditional Artificial Neural Network based Software cost estimation models. For the purpose of evaluation of proposed model, Magnitude of Relative Error and Median of Magnitude of Relative Error are used as a measure of performance index to weigh the obtained quality of estimation which becomes more evident when later compared with two existing models. After an extensive evaluation of results, it showed that the proposed model performs well in software cost estimation.