Enhancing the Software Effort Estimation using Outlier Elimination Methods for Agriculture in Pakistan
Nazish Murtaza, Ahsan Raza Sattar, Tasleem Mustafa · 2010
Introduction Agriculture in Pakistan provides the means of living for more than 75 percent of its population. Any effort to improve crop production necessarily involves steps to manage the water management. It can be achieved using automated water management system. Many software are available for agriculture field but still there is lack of software that are used in precision agriculture; particularly in water management systems. As the progress increases in every field of life, the software project management has also improved. A poor estimate of effort and schedule is often suggested as a major contributor to software project failure. Most of the studies have also paid attention on the development of software effort estimation without consideration of outliers in data sets that cause the wrong results and decisions after implementing these software effort estimation methods. In this paper, we investigated the influence of outlier elimination upon the accuracy of software effort estimation through experiments applying two outlier elimination methods (K-means clustering and My-K-means clustering) and two effort estimation methods( Least squares and Neural network) associatively. This paper proposes a new outlier elimination method My-K-means clustering, which gives better estimation results than K-means clustering. The experiments are performed using the data of Agriculture in Pakistan, with or without outlier elimination. The estimated values of software effort showed the precision of research to improve the automated water management system. The experimental results are favorable because the minimum MMRE is 0.2078 and the maximum Pred (0.25) is 0.7454 using My-K-means clustering.