Recent Trends

In recent years, there has been remarkable development in the software and research industry. Initially, deterministic techniques were used to solve Non-deterministic Polynomial-time NP-hard problems and complex systems. Nowadays, evolutionary algorithms are used to frequently solve complex problems. These algorithms are based on Darwin’s theory of evolution. Chromosomes, the key element in evolutionary algorithms, are used to represent individuals. These chromosomes are fixed- length strings. All these algorithms work in two phases - firstly, a random selection of the individual population is made and secondly, a fitness function is used for selecting the most suitable candidate. Crossover and mutation operations are used to find new generations. It may be stated that evolutionary algorithms use iterative progress, such as population growth. Population selection is made using a guided random search and by parallel processing to find the desired results. All these processes are inspired by the biological evolution mechanism. This evolution has many applications in computer science.

The following are the major evolutionary algorithms: Genetic algorithm. Honeybee Ant Colony optimization, Particle Swarm optimization, Differential Evolution Harmony search, Genetic Programming Cultural algorithm, and Cuckoo search.

One of the emerging research areas is to find a Genetic algorithm contribution to software engineering. Software engineering implies a systematic approach to maintain the system. Using software metrics, we can measure the performance of the software. The main aim of software testing is to find errors or “bugs” present in the software. Software quality assurance is used to judge and improve its quality. Many authors used different techniques to estimate the cost of the software module. Recent advances in the area of component-based software cost estimation are using genetic algorithms to find the software cost.

Exponential growth had been noted in the use of genetic algorithms in recent years. Genetic algorithms are used in heterogeneous fields, e.g., Agriculture, Physics, Mechanical Engineering, Chemistry, Astronomy, Computer Science, Medical Science, etc. The accuracy of the genetic algorithm is excellent.

The following are the research that has been done in this area:

(1) Maleki and Ebrahimi (2014) proposed a method that is a combination of the Firefly algorithm and Genetic algorithm. The NASA 93 dataset is used for the evaluation. In this method, the elitism operation of a Genetic algorithm is used to find the most suitable answer for factors and fitness function of assessment, and this results in a lower error rate. The findings indicate that the average value of the relative error predicted by the COCOMO model is 58.80, and in Genetic and Firefly algorithms is 38.31 and 30.34, respectively, and in the proposed composition model is 22.53.

  • (2) Sharma and Fotedar (2014) proposed a review of different datamining techniques that are useful for effort estimation. One of the important tasks in cost prediction is effort estimation and it is listed under the planning phase of software project management. Some of the methods taken into account are clustered techniques, for example, К-Means, K-NN-K-Nearest Neighbor, Regression technique, Multivariate Analysis for Regression Splines (MARS), Ordinary Least Square regression (OLS), Support Vector Regression (SVR), Classification And Regression Trees (CART); and classification techniques, namely Support Vector Machine (SVM), and Case Based Reasoning (CBR). Hybrid approaches can be used for the above methods for enhancing effort estimation. The proposed technique uses some of the data mining methods that have been detailed to improve the precision of software effort estimation. Effort is being calculated on the basis of the MMRE rate. The lower the MMRE value the technique is supposed to be better. From now onwards, a hybrid approach for any of the datamining methods for increasing the precision in effort estimation can be used. Some of the datasets that can be used are National Aeronautics and Space Administration (NASA), COCOMO 81, International Function Point User Group (IFPUG). Researchers can also use datasets from the COTS project.
  • (3) Morera et al. (2017) validated a genetic framework for effort estimations. The authors also performed a sensitivity analysis for different genetic configurations. Performance results were investigated for the best-learning algorithms. There is much scope for research into the learning schemes used, including 600 learning schemes, eight different processors, five attribute selectors, and 15 modeling techniques. The elitism genetic framework technique is used to automatically select the best learning scheme. A learning scheme is called the best if it has the highest coefficient correlation together with data processing, attribute selection, and learning algorithms. These selected learning schemes are applied to datasets extracted from the ISBSG R12 dataset.

The results show that the performance of the proposed Genetic algorithm is equally good when compared with an exhaustive framework. Sensitivity analysis shows the stability among different genetic configurations. The proposed framework is stable, and performance is better than a random approach. Results show that assembling machine-learning techniques can further optimize efforts estimation.

(4) Krishna and Krishna (2012) proposed a new technique for estimating software costs using Particle Swarm optimization and Fuzzy Logic. Software cost estimation depends upon the size of projects and its proposed parameters. Uncertainty is measured using Fuzzy Logic and Particle Swarm optimization is used for parameters. In these proposed techniques, the triangular membership function of Fuzzy Logic is used. Authors evaluated the proposed method on NASA 63 dataset. Findings show that the proposed technique has a lower rate of error compared to other models.

  • (5) Dizaji and Gharehchopogh (2015) used a hybrid of Ant Colony optimization and Chaos optimization for estimating software cost. To generate random data Lorenz mapping is used for the Chaos optimization algorithm and training is done by using the Ant Colony algorithm. This proposed algorithm is evaluated using NASA 63 dataset. Better performance, as compared to the COCOMO model, is noted in this algorithm in terms of performance and errors.
  • (6) Gharehchopogh and Pourali (2015) scrutinized adjusted COCOMO II model by using a dataset of NASA projects to examine the effect of the developed model and showed the effectiveness of the proposed algorithm in the correction of parameters of COCOMO II. Experiment results show that this model offers an excellent estimate for software cost estimation. In this book, the authors aim to develop an evolutionary model for software cost estimation by using continuous genetic algorithms.

At the onset, a lot of techniques are being used for estimating efforts of CBS using genetic algorithms. Some of these techniques use hybrid approaches with Evolution algorithms, Datamining, Machine Learning, Fuzzy Logic, etc.

The trend of future research is to use genetic algorithms for software effort estimations.


Dizaji, Z. A., & Gharehchopogh. F. S. (2015). A hybrid of ant colony optimization and chaos optimization algorithms approach for software cost estimation. Indian Journal of Science and Technology, 8(2), 128-133.

Gharehchopogh, F. S., & Pourali, A. (2015). A new approach based on continuous genetic algorithm in software cost estimation. Journal of Scientific Research and Development, 2, 87-94.

Krishna, B., & Krishna, T. K. R. (2012). Fuzzy and swarm intelligence for software effort estimation. Advances in Information Technology and Management, 2(1), 246-250.

Maleki, L., & Ebrahimi, F. S. (2014). Gharehchopogh, “A hybrid approach of firefly and genetic algorithms in software cost estimation”. MAGNT Research Report, 2(6), 372-388.

Murillo-Morera, J., Quesada-Lopez, C., Castro-Herrera, C., & Jenkins, M. (2017). A genetic algorithm-based framework for software effort prediction. Journal of Software Engineering Research and Development, 5(1), 4. doi: 10.1186/s40411-017-0037-x.

Sharma, M., & Fotedar, N. (2014). Software effort estimation with data mining techniques- A review. International Journal of Engineering Sciences and Research Technology, pp [1646-1653 3.


Research Question 1

Among all the genetic frameworks find out the best suitable framework for performance as compared to the exhaustive baseline framework.

By framework, we mean generation, population mutation levels, and crossover.

Research Question 2

Find out similarities and differences in the genetic framework for evaluation and prediction phases.

Research Question 3

List the genetic framework learning scheme that is most often selected.

Research Question 4

Find out the best-performing learning scheme reported in terms of evaluation criteria.

< Prev   CONTENTS   Source