Public transportation is important to sustainable mobility in Saudi Arabia, but adoption among female university students remains underexplored. This study examines behavioral intention using an extended Technology Acceptance Model comprising perceived usefulness, perceived ease of use, perceived safety, privacy and comfort, social influence, service quality, perceived affordability, and environmental concern. PLS-SEM was used to examine associations with behavioral intention, while logistic regression, random forest, and support vector machines explored patterns associated with current public-transportation use. The model explained 79.3% of the variance in behavioral intention (R² = 0.793). Perceived safety had the largest positive path coefficient (β = 0.457; f² = 0.283). Logistic regression achieved the highest reported cross-validated ROC-AUC (0.756). The findings are exploratory and require cautious interpretation given the sample and discriminant-validity limitations.
Sustainable forest management requires analytical tools that enable forest managers to evaluate how harvesting decisions affect future forest structure and ecosystem services. An increase in the reliability of predicting how trees will develop has occurred because of the use of models and machine learning techniques over the past several years, however, there are limitations for support management decisions at an operational level. The purpose of this study is to propose a decision-support system that uses artificial intelligence and scenario simulation to develop an intelligent prescription for managing tree selection, applied to long-term inventory data from permanent plots, with the implications visualised in an interactive dashboard. Linear regression modelling yielded the best prediction models, with the root mean square error being 0.988 cm on absolute diameter. Benchmarked against a persistence rule that predicts no change between censuses, which attains a root mean square error of 1.124 cm, the fitted model improves on doing nothing by approximately 15%. On the increment scale, which is the quantity the simulation propagates, the coefficient of determination is 0.174, so accuracy measured on absolute diameter substantially overstates the modelling of growth itself. An interactive dashboard has been developed that provides forest practitioners with tools to aid in both automatic and manual tree selection. The results of the test of the system were evaluated by twelve forest management professionals who compared two separate dashboard designs (DSS A and DSS B). The DSS B interface received a higher mean usability score, although the difference was not statistically significant and the analytical value ratings of the two interfaces were indistinguishable. The evaluation is therefore formative rather than confirmatory, and a counterbalanced task-based study is reported as ongoing work. In conclusion, the results indicate the efficacy of using predictive modelling coupled with scenario simulation to provide forest management professionals with intelligent tools to help make their management decisions.
Organ allocation depends on timely and trustworthy exchange of donor and recipient information among hospitals that run independent information systems. Existing blockchain solutions for healthcare usually rely on a single ledger, which forces a trade-off between the confidentiality required for donor data and the public auditability required for allocation decisions. To resolve this trade-off, this study proposes LifeChain, a hybrid cross-chain framework that keeps donor records on a permissioned Hyperledger Fabric ledger, publishes organ requests and match confirmations on a public Ethereum network, and links the two through a Hyperledger Cactus interoperability gateway. Every cross-chain relay carries a Keccak-256 digest and an Elliptic Curve Digital Signature Algorithm (ECDSA) signature of the gateway, which the Ethereum contract verifies on-chain; the contract thus authenticates the gateway's attestation, while the correctness of the attested Fabric state rests on the gateway, a trust assumption that the study makes explicit. On a single-host prototype with a two-organization Fabric network and the Ethereum Sepolia testnet, under light load, the private ledger processed 153.9 and 154.3 transactions per second (TPS) in a baseline run and a 100-transaction stress run; the public contract interface processed 20.8 and 20.5 TPS and the Cactus bridge 57.8 and 58.8 TPS; these are observed rates, not capacities. Off-chain processing of a cross-chain match took 251.9 ms (baseline run), excluding public block inclusion, which adds at least one 12 s Sepolia slot. All altered and forged relays submitted by the security audit were rejected, and the digests matched on both ledgers for every confirmed match. Compared with a homogeneous Fabric-to-Fabric federation, LifeChain adds public verifiability across heterogeneous ledgers at the cost of a confirmation time dominated by public-chain inclusion.
Continuous Integration and Continuous Delivery pipelines are highly dependent on automated regression tests to enable quick and effective feedback to the developers. Test Case Prioritisation provides a useful approach by arranging the tests in a sequence such that the ones that are likely to yield most faults are run first. Most of the current models are based on heavyweight learners that require significant training time. This article will tackle this problem by introducing a Continuous Integration and Continuous Delivery-ready defect prediction pipeline using LightGBM, which will enable predicting fault-proneness estimates in limited time. The analysis shows that LightGBM can perform equally well or better in predictive accuracy and can significantly decrease training and scoring time. These results make LightGBM-based defect prediction a practical foundation for time-sensitive, risk-aware test prioritisation in Continuous Integration and Continuous Delivery pipelines.
Reliable price estimation supports housing policy and market transparency in Saudi Arabia, yet the online listings that form the main public source of price data are noisy, heavily duplicated, and describe many attributes only in Arabic free text. This study develops and evaluates a reproducible pipeline for predicting apartment prices from 19,971 for-sale listings scraped from the Aqar.fm platform. A documented cleaning procedure removed 25.2% of the raw records, chiefly repeated advertisements, and retained 14,934 listings concentrated in Jeddah, Riyadh, and Dammam. Fifty binary features were derived from the Arabic descriptions with spelling-variant-aware keyword rules. Multiple linear regression, ridge regression, random forest, LightGBM, and XGBoost were compared on log price under five-fold grouped cross-validation, which keeps listings sharing a description group out of the opposite partition, with hyperparameters selected over widened grids. LightGBM was selected by the cross-validation criterion and outperformed the linear baselines on every test-set metric at both the log-price and price scale: a mean absolute error of 69,985 SAR (95% bootstrap CI 65,005–75,683) and MAPE of 9.3%, against 93,161 SAR and 12.9% for multiple regression. Location features accounted for most of the gain over structural attributes alone (MAE 134,218 to 73,645 SAR), Arabic text describing the building added a further 4.9% reduction (to 70,016 SAR), and neighbourhood-service text added no further reduction in error. A model trained on only two of the three cities performed poorly when tested on the held-out city (MAPE 20.2% versus 7.0% under a random split; R² −0.28), a result attributable mainly to location-identity encodings that are undefined for an unseen city rather than to a general failure of transfer. Permutation importance, grouped permutation importance, and TreeSHAP analyses are reported to explain the selected model's predictions, which are interpreted as predictive associations rather than causal effects.
