Graph based transfer learning with orthogonal tunning for functionality size insights

This study proposes a Function Point Analysis (FPA) based software development effort estimation methodology integrated with Use Case Diagrams. These methods include identifying actor activities, classifying those activities into FPA categories, and calculating Unadjusted Function Points (UFP). Followed by the calculation of Technical Complexity Factors (TCF) and Adjusted Function Points (AFP), this study aims to produce more accurate man-hours estimates. Results show a UFP of 162 TCF of 11, AFP of 123.12, and an estimated effort of 1846.8 hours worked, while the actual effort is 1228 hours. Evaluation of estimates using the metrics Mean Magnitude of Relative Error (MMER) 0.34, Mean Magnitude of Relative Error (MMRE) 0.50, Mean Absolute Error (MAE) 618.80, Mean Balanced Relative Error (MBRE) 0.50, Mean Inverse Balanced Relative Error (MIBRE) 0.34, and Root Mean Squared Error (RMSE) 618.80, showed sufficient precision despite the overestimation. The study suggests the need for adjustments in TCF calculations and considering development environment factors in more detail to improve estimation accuracy. These findings are essential in improving the precision of effort estimation methodologies in software development, particularly in projects that use Use Case Diagrams as the primary framework.

This paper proposes to establish a framework for use in the earliest stages of a software project for comparing the various types of software development, thus allowing this information to be part of the project selection process. The possible types of software development projects are classified into three categories: new development, reuse-based without modification, and reuse-based with modification. We propose to utilize a decision tree framework that includes quantitative risk management related to project cost and possible project success. The quantitative risk analysis consists of cost estimation, possibility (success rate), Net Present Value (NPV), and Expected Monetary Value (EMV). Probabilistic schedule estimation based on the International Function Point User Group (IFPUG) function point analysis and the International Software Benchmarking Standards Group (ISBSG) regression equations (on which we previously conducted research) is also fundamental to the project cost estimation of each decision node. The use of the proposed framework provides more accurate and objective comparisons among the types of software development in the early stages of the software project life cycle.

The n-octanol/buffer solution distribution coefficient at pH = 7.4 (log D7.4) is an indicator of lipophilicity, and it influences a wide variety of absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties and druggability of compounds. In log D7.4 prediction, graph neural networks (GNNs) can uncover subtle structure-property relationships (SPRs) by automatically extracting features from molecular graphs that facilitate the learning of SPRs, but their performances are often limited by the small size of available datasets. Herein, we present a transfer learning strategy called pretraining on computational data and then fine-tuning on experimental data (PCFE) to fully exploit the predictive potential of GNNs. PCFE works by pretraining a GNN model on 1.71 million computational log D data (low-fidelity data) and then fine-tuning it on 19,155 experimental log D7.4 data (high-fidelity data). The experiments for three GNN architectures (graph convolutional network (GCN), graph attention network (GAT), and Attentive FP) demonstrated the effectiveness of PCFE in improving GNNs for log D7.4 predictions. Moreover, the optimal PCFE-trained GNN model (cx-Attentive FP, Rtest2 = 0.909) outperformed four excellent descriptor-based models (random forest (RF), gradient boosting (GB), support vector machine (SVM), and extreme gradient boosting (XGBoost)). The robustness of the cx-Attentive FP model was also confirmed by evaluating the models with different training data sizes and dataset splitting strategies. Therefore, we developed a webserver and defined the applicability domain for this model. The webserver (http://tools.scbdd.com/chemlogd/) provides free log D7.4 prediction services. In addition, the important descriptors for log D7.4 were detected by the Shapley additive explanations (SHAP) method, and the most relevant substructures of log D7.4 were identified by the attention mechanism. Finally, the matched molecular pair analysis (MMPA) was performed to summarize the contributions of common chemical substituents to log D7.4, including a variety of hydrocarbon groups, halogen groups, heteroatoms, and polar groups. In conclusion, we believe that the cx-Attentive FP model can serve as a reliable tool to predict log D7.4 and hope that pretraining on low-fidelity data can help GNNs make accurate predictions of other endpoints in drug discovery.

Understanding and interpreting machine learning models is crucial in high-stakes industries like steel manufacturing, where decisions impact energy efficiency and environmental sustainability. Traditional feature importance methods often struggle with robustness under noisy conditions, leading to unreliable insights. To address this problem, we introduce the Complexity and Interaction Enhanced SHAP (CESHAP), a novel feature importance method that incorporates model complexity and feature interactions. Alongside, we propose the Noise-Adjusted Feature Importance Change (NAFIC) metric to assess the robustness of feature importance methods against varying levels of noise. Experiments conducted on an energy consumption dataset from the steel industry, with systematically introduced Gaussian noise levels (5%, 10%, 15%, 20%), demonstrate that CESHAP offers superior robustness in tree-based models compared to SHapley Additive exPlanations (SHAP) and Permutation Feature Importance (PFI). Our findings underscore the effectiveness of CESHAP in enhancing interpretability and reliability in complex, non-linear models, ultimately supporting more informed decision-making in energy-intensive industries.

A fuzzy logic based approach for phase-wise software defects prediction using software metrics

Age is a significant risk factor for mild cognitive impairment (MCI) and Alzheimer’s disease (AD) and identifying brain age patterns is critical for comprehending the normal aging and MCI/AD processes. Prior studies have widely established the univariate relationships between brain regions and age, while multivariate associations remain largely unexplored. Herein, various artificial intelligence (AI) models were used to perform brain age prediction using an MRI dataset (n = 825). The optimal AI model was then integrated with the feature importance methods, namely Shapley additive explanations (SHAP), local interpretable model-agnostic explanations, and layer-wise relevance propagation, to identify the significant multivariate brain regions hierarchically involved in this prediction. Our results showed that the deep learning model (referred to as AgeNet) outperformed conventional machine learning models for brain age prediction, and that AgeNet integrated with SHAP (referred to as AgeNet-SHAP) identified all ground-truth perturbed regions as key predictors of brain age in semi-simulation, demonstrating the validity of our methodology. In the experimental dataset, when compared to cognitively normal (CN) participants, MCI exhibited moderate differences in brain regions, whereas AD showed highly robust and widely distributed regional differences. Individualized AgeNet-SHAP regional features further showed associations with clinical severity scores in the AD continuum. These results collectively facilitate data-driven explainable AI approaches for disease progression, diagnostics, prognostics, and personalized medicine efforts.

Urban morphology plays a pivotal role in shaping urban heat islands (UHI), especially in high-density cities, by influencing land surface temperature (LST) and air temperature. While previous studies have explored the relationships between urban morphology and temperature, they often fail to simultaneously capture the nonlinear interactions of these relationships and spatial heterogeneity arising from regional variations in urban form and environmental conditions. Furthermore, the varying influence of urban morphology on diurnal and nocturnal UHI remains insufficiently understood. This study bridges this gap by applying Geographically Weighted Machine Learning (GWML) and Graph Neural Network (GNN) models to investigate the non-stationary relationships between UHI and its influencing factors. Using high-resolution diurnal and nocturnal LST data from Landsat and ECOSTRESS, combined with 3D building morphology metrics (e.g., Sky View Factor), road network attributes, socio-demographic characteristics, and landscape indices, we systematically analyse the spatial variations in these associations. The analysis includes calculating Moran’s I to detect spatial patterns and comparing the predictive performance of GWML and GNN against Geographically Weighted Regression (GWR). SHapley Additive exPlanations (SHAP) enhance interpretability of explainable GeoAI (X-GeoAI) models, revealing localized impacts of key influencing factors. Our findings demonstrate significant spatial variations in the effects of 2/3D urban morphology on UHI across diurnal and nocturnal cycles. These insights provide a robust foundation for targeted UHI mitigation strategies and adaptive urban planning. This work highlights the potential of advanced GeoAI methods in urban climate research and offers actionable pathways for enhancing climate resilience in high-density cities.

Abstract. As air pollution is regarded as the single largest environmental health risk in Europe it is important that communication to the public is up to date and accurate and provides means to avoid exposure to high air pollution levels. Long- and short-term exposure to outdoor air pollution is associated with increased risks of mortality and morbidity. Up-to-date information on present and coming days' air quality helps people avoid exposure during episodes with high levels of air pollution. Air quality forecasts can be based on deterministic dispersion modelling, but to be accurate this requires detailed information on future emissions, meteorological conditions and process-oriented dispersion modelling. In this paper, we apply different machine learning (ML) algorithms – random forest (RF), extreme gradient boosting (XGB), and long short-term memory (LSTM) – to improve 1, 2, and 3 d deterministic forecasts of PM10, NOx, and O3 at different sites in Greater Stockholm, Sweden. It is shown that the deterministic forecasts can be significantly improved using the ML models but that the degree of improvement of the deterministic forecasts depends more on pollutant and site than on what ML algorithm is applied. Also, four feature importance methods, namely the mean decrease in impurity (MDI) method, permutation method, gradient-based method, and Shapley additive explanations (SHAP) method, are utilized to identify significant features that are common and robust across all models and methods for a pollutant. Deterministic forecasts of PM10 are improved by the ML models through the input of lagged measurements and Julian day partly reflecting seasonal variations not properly parameterized in the deterministic forecasts. A systematic discrepancy by the deterministic forecasts in the diurnal cycle of NOx is removed by the ML models considering lagged measurements and calendar data like hour and weekday, reflecting the influence of local traffic emissions. For O3 at the urban background site, the local photochemistry is not properly accounted for by the relatively coarse Copernicus Atmosphere Monitoring Service ensemble model (CAMS) used here for forecasting O3 but is compensated for using the ML models by taking lagged measurements into account. Through multiple repetitions of the training process, the resulting ML models achieved improvements for all sites and pollutants. For NOx at street canyon sites, mean squared error (MSE) decreased by up to 60 %, and seven metrics, such as R2 and mean absolute percentage error (MAPE), exhibited consistent results. The prediction of PM10 is improved significantly at the urban background site, whereas the ML models at street sites have difficulty capturing more information. The prediction accuracy of O3 also modestly increased, with differences between metrics. Further work is needed to reduce deviations between model results and measurements for short periods with relatively high concentrations (peaks) at the street canyon sites. Such peaks can be due to a combination of non-typical emissions and unfavourable meteorological conditions, which are rather difficult to forecast. Furthermore, we show that general models trained using data from selected street sites can improve the deterministic forecasts of NOx at the station not involved in model training. For PM10 this was only possible using more complex LSTM models. An important aspect to consider when choosing ML algorithms is the computational requirements for training the models in the deployment of the system. Tree-based models (RF and XGB) require fewer computational resources and yield comparable performance in comparison to LSTM. Therefore, tree-based models are now implemented operationally in the forecasts of air pollution and health risks in Stockholm. Nevertheless, there is big potential to develop generic models using advanced ML to take into account not only local temporal variation but also spatial variation at different stations.

Function Point Analysis (FPA) is a largely used technique to estimate the size of development project, enhancement project or applications already installed. During the point counting process that represents the dimension of a project or an application, each function is classified according to its relative functional complexity. Several studies resulted in FPA extensions, and most of them are mainly aimed at achieving greater precision in the point assessment of systems of greater algorithmic complexity. This work proposes the use of concepts and properties from fuzzy set theory to extend FPA into FFPA (Fuzzy Function Point Analysis). Fuzzy theory seeks to build a formal quantitative structure capable of emulating the imprecision of human knowledge. With the function points generated by FFPA, the functionality of the project is better represented than it was through FPA. Thus, derived values such as costs and terms of development can be more precisely determined.

In this study, we extend the study of Function Point Analysis (FPA) for the measurement of source code based migration project using the migration tooling options. FPA is an objective and structured technique to measure software size by quantifying its functionality provided to the user, based on the requirements and logical design. It is noticed that the procedure for the above measurement is not fully described in FPA method. Hence, the same estimation concept has been applied to an in-house project on an experimental basis at a leading software development organization. The yielded results are compared with FPA for the re-engineering project.

<i>Introduction</i>: The precise estimation of software effort is a significant difficulty that project managers encounter during software development. Inaccurate forecasting leads to either overestimating or underestimating software effort, which can be detrimental for stakeholders. The International Function Point Users Group&#x2019;s Function Point Analysis (FPA) method is one of the most critical methods for software effort estimation. However, the practice of using the FPA method in the same fashion across all software areas needs to be reexamined. <i>Aim</i>: We propose a model for evaluating the influence of data clustering on software development effort estimation and then finding the best clustering method. We call this model the effort estimation using machine learning applied to the clusters (EEAC) model. <i>Method</i>: We cluster the dataset according to the clustering method and then apply the FPA and EEAC methods to these clusters for effort estimation. The clustering methods we use in this study include five categorical variable criteria (Development Platform, Industrial Sector, Language Type, Organization Type, and Relative Size) and the k-means clustering algorithm. <i>Results</i>: The experimental results show that the estimation accuracy obtaining with clustering consistently outperforms the accuracy without clustering for both the FPA and EEAC methods. Significantly, using the FPA method, the average improvement rate from using clustering as opposed to non-clustered was highest at 58.06%, according to the RMSE. With the EEAC method, this number reached 65.53%. The Industry Sector categorical variable achieves the best accuracy estimation compared to the other clustering criteria and k-means clustering. The improvement in accuracy in terms of the RMSE when applying this criterion is 63.68% for the FPA method and 72.02% for the EEAC method. <i>Conclusion</i>: Better results are obtained through dataset clustering compared to no clustering for both the FPA and EEAC methods. The Industry Sector is the most suitable clustering method among the tested clustering methods.

In this study, we present a window-based exponential effort estimation model to predict the effort required in terms of man days by using Unadjusted Function Point (UFP) size measure and eliminate the usage of General Systems Characteristics (GSCs). A very comprehensive statistical analysis and test was carried out on large amount of quality project data in the International Software Benchmarking Standard Group (ISBSG) Release 9 dataset, which were collected by the International Function Points User Group (IFPUG) count approach in the process of model development. The effectiveness of the model was examined and reported in this study.

Service-Oriented Architecture (SOA) is an architectural concept to build software which emphasizes the use of services and comprises design principles. Research on this paper implements SOA design principles in the development of Animal Video Encyclopedia and Plant Video Encyclopedia. By implementing SOA, this paper would like to see if SOA affects the number of Function Points in the development of a software solution. Function Point numbers which are the result of Function Point Analysis (FPA) calculation have been widely used in software industry to measure the size of an application based on functionalities from user’s perspective. This paper uses FPA method to count the Function Points between both Animal and Plant Video Encyclodias solution implementing SOA and Animal and Plant Video Encyclopedias solution that does not implement SOA. Function Points from both solutions are then compared and analyzed based on the implementation of SOA design principles. Results show that numbers of Function Points of the solution implementing SOA appear to be smaller than the number of Function Points of the solution which is not implementing SOA.

This paper investigates correlating the basic elements of Unified Modeling Language and Cyclomatic Complexity with Function Point Analysis (FPA) principles to develop an automated software functional sizing tool. This concept has been difficult to achieve due to the logical nature of the FPA sizing methodology versus the physical nature of source lines of code (SLOC). In this approach, we examine software complexity from design and maintainability perspectives in order to understand relationships in physical code. Our hypothesis is that this method will “simulate” FPA principles and produce an objective sizing method. This would provide the foundation for an automated tool that scans physical software code to derive “Objective Function Points”(OFPs) functional size measure.

To maintain customer trust, Islamic banking continues to improve its services which rely on the use of information technology (software and its supporting peripherals). In the meantime, research on software in Islamic banking are still finite. Thus, the authors are interested in calculating the size and effort of working on the core application of an Islamic bank, which is called the Islamic Banking Application (IBA). Two methods are used in the study to see the size of the IBA. The size can be seen from the point of view of functional-based processes and non-functional aspects. The functional one, we propose the Use Case Points (UCP) method. In the UCP method, there are at least4 steps that must be passed for measuring software size. On the other hand, for non-functional measurement, we used the original Function Points Analysis (FPA) concepts. As a result, this research produces size measurement software using the FPA technique of 2,109.40 function points, while the UCP method results in 3,024.54 use case points. To convert size measurement into effort estimation, the author uses previous research with a productivity rate of 8.2 person/hour. Thus, the predicted effort consumed to develop the IBA is 17,297.08 person/hour (based on FPA technique) and 24, S01.23 person/hour (by UCP method). We found the deviation of 7,504 person/hour between both methods. This is due to the existence of different measurement parameters.