The average software company spends a huge amount of its revenue on R&D for how to deliver software on time. Accurate software effort estimation is critical for successful project planning, resource allocation, and on-time delivery within budget for sustainable software development. However, both overestimation and underestimation pose significant challenges in software development, necessitating continuous improvement in estimation techniques. This study reviews recent machine learning approaches exploited to enhance software effort estimation (SEE) accuracy, focusing on research published between 2020 and 2023. The literature review employed an approach to identify pertinent research on machine learning techniques for software estimation efforts. Additionally, comparative experiments were conducted employing five commonly used ML methods: K-Nearest Neighbor, Support Vector Machine, Random Forest, Logistic Regression, and LASSO Regression. These techniques were assessed using five widely employed accuracy metrics such as Mean Squared Error (MSE), Mean Magnitude of Relative Error (MMRE), R-squared, Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE) on seven benchmark datasets (Albrecht, Desharnais, China, Kemerer, Mayazaki94, Maxwell, COCOMO). By carefully reviewing study quality, analyzing results across the literature, and rigorously evaluating experimental outcomes, clear conclusions were drawn about the most promising techniques for achieving state-of-the-art accuracy in estimating software effort. This study makes three key contributions to the field: firstly, it furnishes a thorough overview of recent machine learning research in software effort estimation (SEE); secondly, it provides data-driven guidance for researchers and practitioners to select optimal methods for accurate effort estimation; and thirdly, it demonstrates the performance of publicly available datasets through experimental analysis. Enhanced estimation supports the development of better predictive models for software project time, cost, and staffing needs. The findings aim to focus future research directions and tool development toward the most accurate machine learning approaches for modeling software development effort, costs, and delivery schedules.

Emperor Penguin Optimizer (EPO) is a recently developed population-based meta-heuristic algorithm that simulates the huddling behavior of emperor penguins. Mixed results have been observed on the performance of EPO in solving general optimization problems. Within the EPO, two parameters need to be tuned (namely f and l) to ensure a good balance between exploration (i.e., roaming unknown locations) and exploitation (i.e., manipulating the current known best). Since the search contour varies depending on the optimization problem, the tuning of f and l is problem-dependent, and there is no one-size-fits-all approach. To alleviate these problems, an adaptive mechanism can be introduced in EPO. This paper proposes a fuzzy adaptive variant of EPO, namely Fuzzy Adaptive Emperor Penguin Optimizer (FAEPO), to solve this problem. As the name suggests, FAEPO can adaptively tune the parameters f and l throughout the search based on three measures (i.e., quality, success rate, and diversity of the current search) via fuzzy decisions. A test suite of twelve optimization benchmark test functions and three global optimization problems (Team Formation Optimization-TFO, Low Autocorrelation Binary Sequence-LABS, and Modified Condition/Decision Coverage-MC/DC test case generation) were solved using the proposed algorithm. The respective solution results of the benchmark meta-heuristic algorithms were compared. The experimental results demonstrate that FAEPO significantly improved the performance of its predecessor (EPO) and gives superior performance against the competing meta-heuristic algorithms including an improved variant of EPO (IEPO).

Test redundancy occurs when one requirement is covered by more than one test. Potentially affecting the testing costs while at the same time delaying the software release, test redundancy is often undesirable. Many recent works have dealt with the test redundancy reduction problem as an optimization problem. Consequently, a plethora of work utilizing meta-heuristic algorithms as the backbone algorithm for addressing the test redundancy reduction problem can be seen in the literature. Although useful, many existing meta-heuristic-based algorithms have focused on solving the test redundancy reduction problem as a single task problem (i.e., one-test redundancy task at-a-time). To cater for simultaneous test redundancy reduction from multiple software development projects, a multi-task-based test redundancy reduction algorithm is desirable. This paper explores the design and implementation of a multi-task sine cosine algorithm (MT_SCA) for test redundancy reduction optimization. More precisely, MT_SCA exploits a population division-based approach to achieve multi-task capability. Experimental results demonstrate that MT_SCA gives comparable test reduction size and execution time against its single solution and other metaheuristic counterparts.

Forming a group of experts who fulfill the requirements of a collaborative endeavor is vital. Aiming to reduce communication costs among the team of experts in their effort to resolve the team formation challenge, this work introduces an enhanced sine cosine algorithm. Improving Sine Cosine Algorithm with a Swap Operator (ESCASO) is the designation given to the proposed method. To augment the diversity of SCA, we employed a swap operator within the sine cosine algorithm. In contrast to other meta-heuristic algorithms such as the grey wolf optimizer, harris hawks optimization algorithm, and traditional sine cosine algorithm, we are evaluating the ESCASO’s performance by applying it to the IMDB dataset. Based on the experimental findings, it can be concluded that the proposed algorithm effectively identifies teams that achieve the goal with minimal communication cost among team’s members.

Background/Objectives: Drug–drug interactions (DDIs) are major medication-safety concerns, and experimental testing cannot cover the expanding number of drug pairs. This systematic review evaluates graph-based machine-learning methods for DDI prediction, focusing on machine-learning architectures, data integration, interpretability, reproducibility, and clinical relevance. Methods: Following PRISMA 2020, we systematically searched major databases for studies published between January 2021 and March 2026. We included studies that applied graph-based machine-learning models, particularly graph neural networks, to predict DDIs. We compared their data sources, model designs, validation methods, predictive performance, reproducibility, and clinical relevance. Because the studies used different datasets and evaluation methods, the findings were summarized narratively rather than combined statistically. Results: Sixty studies met the eligibility criteria. Methods progressed from graph convolutional networks and graph attention networks to graph transformers, contrastive learning, multimodal fusion, and LLM-enhanced representations. We found that reported improvements in prediction performance often remained study-specific. Only three studies explicitly mentioned or addressed data leakage, whereas most reviewed studies contained no explicit leakage discussion; leakage-aware drug-disjoint, temporal, and external evaluations were also uncommon. Uncertainty calibration, computational-resource reporting, complete reproducibility materials, and independently validated explanations were also limited. Conclusions: Graph-based machine learning is promising for DDI prioritization and hypothesis generation but remains insufficient for independent clinical decision-making. Future studies should use standardized benchmarks, leakage-aware validation, calibrated uncertainty, reproducible pipelines, validated explanations, and external or prospective evaluation.