ABSTRACT Software effort estimation (SEE) determines the effort necessary to develop software. The researchers have been tending to SEE issues since the 1960s, and several methods have been created until the formulation of the function point (FP) and constructive cost estimation (COCOMO) methods. However, these methods are only useful for procedurally developed software, not modern object‐oriented (OO) software. Because the use case is the widely used unit of an OO system, particularly in scenarios requiring structured and early‐stage effort estimation, using the use case point (UCP) approach will help get accurate results. The UCP approach consists of size estimation (in UCP) and effort estimation with calculated size. This study focuses on effort estimation when the size (in UCP) is already known. The productivity of a project is one of the main components for estimating effort from the given size. The classical SEE models based on UCP utilized a fixed number of productivity values. So, the validity of classical approaches is a subject of disapproval because of static productivity values. Purposefully, we proposed a rank‐based weighted ensemble model for productivity prediction that allows us to use flexible productivity values. We used learning techniques such as simple linear regression (SLR), Least Absolute Shrinkage and Selection Operator Regression (LR), ridge regression (RR), elastic net regression (ER), K‐nearest neighbor (KNN), decision tree (DT), support vector regression (SVR), multilayer perceptron (MLP), bagging, and adaptive boosting for productivity prediction and compared them with the proposed model. Further, we used existing UCP prediction models and compared the proposed approach with them.

Software Effort Estimation (SEE) is one of the most challenging tasks in software project management. In the literature, the researchers addressed the SEE problem in different ways, including models developed using machine learning (ML). Integrating individual models (Ensemble) is an active research topic in the ML domain, leading to improved performance than the individual models. Ensembling approaches such as bagging, boosting, and stacking have been extensively studied in the literature for improved SEE. The performance of an ensemble model largely depends upon tuning the hyperparameters of each learner and assigning the correct weight to each base learner. To this end, this work proposes a self-adaptive ensemble-based approach that combines hyperparameter tuning and weight assignment problems in a single step and solves them using optimization problems. The proposed model optimizes the hyperparameters of each learner using the particle swarm optimization (PSO) algorithm. Also, the weights of each base learner are optimized using Sequential Least SQuares Programming (SLSQP). An experimental analysis is conducted to evaluate the proposed model over different SEE datasets from PROMISE and ISBSG data repositories. The results show that the proposed ensemble approach performed better than the individual models. Further, we compared our proposed approach with other commonly used ensembling approaches in the literature. We found that the proposed ensemble approach provides better predictive performance than other ensembles.

Effort Estimation is a very challenging task in the software development life cycle. Inaccurate estimations may cause the client dissatisfaction and thereby, decrease the quality of the product. Considering the problem of software cost and effort prediction, it is conceivable to call attention to that the estimation procedure considers the qualities present in the data set, as well as the aspects of the environment in which the model is embedded. Existing literatures have the instances where machine learning techniques such as Linear Regression (LR), Support Vector Machine (SVM), K-Nearest Neighbor (KNN) have been used to estimate the effort required to develop any software. Yet it is quite uncertain for any particular model to perform well with all the data sets. Most of the research is based on the dataset of any single organization. Consequently, the results obtained through these models cannot be generalized. So, the main objectives of this research are: i) to use different data preparation techniques such as selection, cleaning, and transformation to improve the quality of data set given to the model ii) to use other machine learning models such as Multi-Layer Perceptron Neural Network (MLPNN), Probabilistic Neural Network (PNN), and Recurrent Neural Network (RNN) to increase the performance of software effort estimation process iii) to use different optimization techniques to tune the parameters of machine learning models iv) to use ensemble methods to improve the accuracy of software effort estimation process. In this study, first, we found out the most influential attributes in the Desharnais data set, then, MLPNN has been applied on reduced data set with to improve the accuracy of software effort estimation. Then, the performance of the MLPNN model is compared with LR, SVM and KNN models in the literature to find the best model fitting this dataset. Results obtained from the study demonstrate that some of the variables are more important in comparison to others for effort estimation. Also among the various models used in this study, the best-obtained R2 value is 79 % for the MLPNN model.

Effort Estimation is a very challenging task in the software development life cycle. Inaccurate estimations may cause client dissatisfaction and thereby, decrease the quality of the product. Considering the problem of software cost and effort estimation, it is conceivable to call attention to that the estimation procedure considers the qualities present in the data set, as well as the aspects of the environment in which the model is embedded. Existing literature have the instances where machine learning techniques have been used to estimate the effort required to develop any software. Yet it is quite uncertain for any particular model to perform well with all the data sets. In this paper, Multi-Layer Perceptron (MLPNN) and its ensembles are explored in order to improve the performance of software effort estimation process. Firstly, MLPNN, Ridge-MLPNN, Lasso-MLPNN, Bagging-MLPNN, and AdaBoost-MLPNN models are developed and, then, the performance of these models are compared on the basis of R2score to find the best model fitting this dataset. Results obtained from the study demonstrate that the R2score of AdaBoost-MLPNN is 82.213%, which is highest among all the models.