A predictive approach to estimate software defects density using Probabilistic Neural Networks for the given Software Metrics

Abstract

Software plays a very important role in our everyday life.There are many instances which show that even a small defect in the software can cause huge loss to many lives.By evaluating the errors in software in various phases,we can reduce the lateral cost of software.Software quality prediction has been an important arena since the las two decades. Several models and techniques have been proposed and utilized in this gaze.We can recognize the areas which are prone to hazards with the help of logic of quality prediction. In the proposed model, defect density indicator in requirement analysis, design, coding and testing phase is predicted using ten software metrics of these four phases.at the end of each phase the defect density indicator will be taken as an input for the next phase.With the help of ANN and PNN strategies,we have extnded our work. The experimental results are compared with fuzzy, ANN and PNN.Incomparision with ANN and fuzzy,the number of defects can be discovered better with ANN strategy and with PNN strategy there is better prediction in the reliability of mertics of SDLC.Compared to the two implementation methods used on all datasets,ProbabilisticNueral Networks have better reliability.Rather than depending on a singlrtechnique,it would be better to use a scope of software defect prediction models.Experimental results will be performed by utilizing matlab tool.

Authors and Affiliations

T. Ravi Kumar, Dr. T. Srinivasa Rao, Dr. Ch. V. M. K. Hari

Keywords

Related Articles

Synthesis, characterization of Banana / Glass Fiber Reinforced Epoxy Based Hybrid Composites

Natural fiber composites are nowadays being used in various engineering applications to increase the strength and optimize the weight and the cost of the product. Hybridization is a process of incorporating synthetic fib...

Phyto cover for Sanitary Landfill Sites: A brief review

Landfill gases (LFG) are produced due to biodegradation of organic fraction of municipal solid waste (MSW) when water comes in contact with buried wastes. The conventional clay cover is still practiced to mitigate the pe...

ANFIS based Doubly Fed Induction Generator for WECS with LLCL-2TRP Filter

The wind energy conversion system integrated with a doubly fed induction generator. The PWM based back-toback power converters produced harmonic and as well as the non-linear loads are also a contributor of harmonics. Fi...

A 0.6-V 2-nA CMOS Current Reference Circuit

This paper proposes a 0.6-V current reference circuit for use in ultra-low-power applications. In a conventional resistorlessbeta-multiplier current reference circuit, a MOSFET that operates in the strong inversion and t...

Optimal Advertisement Placement Slot Using Knapsack Problem (A Case Study of Television Advertisement of Tv 3 Ghana)

It is undoubtedly that the major goal of many profit making organizations is to maximize their profits. And this is achieved by ensuring that some parameters such as time, materials, employees, among others are managed p...

Download PDF file
  • EP ID EP394952
  • DOI 10.9790/9622-0807020815.
  • Views 119
  • Downloads 0

How To Cite

T. Ravi Kumar, Dr. T. Srinivasa Rao, Dr. Ch. V. M. K. Hari (2018). A predictive approach to estimate software defects density using Probabilistic Neural Networks for the given Software Metrics. International Journal of engineering Research and Applications, 8(7), 8-15. https://europub.co.uk/articles/-A-394952