Comparing some neural network models for software development effort prediction

• Published in: 2011 2nd National Conference on Emerging Trends and Applications in Computer Science pp. 1 - 4
• Main Authors: Ghose, M K, Bhatnagar, R, Bhattacharjee, V
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.03.2011
• Subjects: Accuracy; Artificial Neural Network; Artificial neural networks; Computational modeling; Effort Estimation; Estimation; Mathematical model; Programming; Soft Computing; Software
• ISBN: 1424495784; 9781424495788
• DOI: 10.1109/NCETACS.2011.5751391

Software development effort estimation is one of the most major activities in software project management. A number of models have been proposed to make software effort estimations but still no single model can predict the effort accurately. The need for accurate effort estimation in software industry is still a challenge. Artificial Neural Network models can be used for carrying out the effort estimations for developing a software project & this field of Soft Computing is suitable in effort estimations. The present paper is concerned with comparing the results of various artificial neural network models for predicting the software development effort estimation. The neural network models available in MATLAB neural network tools were used and the standard dataset as compiled by Lorenz et.al. was used in the present study. The results were analyzed using four different criterions MRE, MMRE, BRE and Pred. It is observed that the Generalised Regression Neural Network model provided better results.