Authors:-Â Anuradha Goel, V. K Laxminarayan
Abstract:-Â It takes estimation of model parameters to forecast a system's behaviour. The estimation of model parameters is done using optimization criteria. Most algorithms estimate model parameters using past data. The output generated by the model is compared to the known goal values (actual). The parameters are estimated based on the differences between the two. Different criteria are applied to compare various models created using the same data. Here, the information gathered for small-scale initiatives is utilised. Using a radial basis function network, we explore the challenge of estimating the software effort. Various current criteria for one and two predictors are used to compare accuracy. Then, we compared the outcomes of one and two predictors and suggested a new assessment criterion based on linear least squares. We have taken into account a different data set and assessed prediction accuracy with the new standard. Compared to a single statistic, the new criteria are simpler to understand. This approach may be used for any modelling and prediction, even though software effort estimation is taken into consideration.
Keywords:-Â Radial Basis Function, Software effort estimation, accuracy, linear least squares.
Â
Full Issue
| View or download the full issue | PDF 97-107 |