Forest Model Research Articles

Research on Prediction of Online Training Effect of Newly Recruited PE Teachers in Junior and Senior Middle Schools Based on Machine Learning

With the continuous development of online education, how to improve the teaching ability of new PE teachers through adaptive and effective online training has become an important research issue. Based on machine learning algorithm, this paper discusses the influence of different characteristics on the adaptability of online training for new physical education teachers, and evaluates the application of various models in predicting the training effect of teachers. The results show that factors such as teachers' professional experience, past training experience, school type and whether to participate in training with colleagues have a significant impact on the adaptability of online training. By comparing Logistic regression model, KNN model, random forest model, XGBoost model and support vector machine model, this paper finds that random forest model is the best in prediction accuracy and generalization ability. This study provides data support and theoretical basis for optimizing the online training of physical education teachers, and can provide reference for educational managers to formulate personalized training programs.

Prediction of potential occurrence of historical objects with defensive function in Slovakia using machine learning approach

In this article, we aim at the prediction of possible locations of already defunct historical objects with a defensive function (HODFs) in Slovakia, which have not been found and documented so far, using three machine learning methods. Specifically, we used the support vector machine, k-nearest neighbors, and random forest algorithms, which were trained based on the following five factors influencing the possible occurrence of HODFs: elevation, distance from a river, distance from a settlement, lithological rock type, and type of representative geoecosystems. Training and testing datasets were based on a database of already documented 605 HODFs, which were divided into 70% of training samples and 30% of testing samples. All of the three models reached the AUC-ROC value over 0.74 based on the testing dataset. The best performance was recorded by the random forest predictive model with the AUC-ROC value equal to 0.79. The results of the random forest model were also validated with the recently documented HODFs via the archeological research.

Gridded global dataset of industrial water use predicted using the Random Forest

Spatially distributed industrial water use (IWU) data are essential for effective region-specific water resource management. Such data are often scarce in underdeveloped and developing countries. We propose a random forest regression model to predict IWU at a spatial resolution of 0.5° by combining socioeconomic, climatic, and geographical datasets. These datasets included nighttime light (NL), global power plants, country-wise IWU, elevation data (DEM), gross domestic product (GDP), road density (RD), cropland (CRP), wetland (WLND), population (POP), precipitation (PCP), temperature (TEMP), wet days (WET) per year, and potential evapotranspiration (PET). The results show that RD, CRP, POP, GDP, DEM, and TEMP were the most influential variables. We assessed the accuracy of the global IWU map using published and observed datasets from various sources for the major industrialized countries such as the USA and China from 2000 to 2015. The predicted global map shows a reasonable distribution of grid-wise values for highly industrialized countries and data-scarce regions. Thus, fine-resolution maps can support local planning and decision-making for large basins worldwide.

Analysis of the Trends and Driving Factors of Cultivated Land Utilization Efficiency in Henan Province from 2000 to 2020

Amid persistent global food security challenges, the efficient utilization of cultivated land resources has become increasingly critical, as optimizing Cultivated Land Utilization Efficiency (CLUE) is paramount to ensuring food supply. This study introduced a cultivated land utilization index (CLUI) based on Fractional Vegetation Cover (FVC) to assess the spatiotemporal variations in Henan Province’s CLUE. The Theil–Sen slope and the Mann–Kendall test were used to analyze the spatiotemporal variations of CLUE in Henan Province from 2000 to 2020. Additionally, we used a genetic algorithm optimized Artificial Neural Network (ANN) and a particle swarm optimization-based Random Forest (RF) model to assess the comprehensive in-fluence between topography, climate, and human activities on CLUE, in which incorporating Shapley Additive Explanations (SHAP) values. The results reveal the following: (1) From 2000 to 2020, the CLUE in Henan province showed an overall upward trend, with strong spatial heterogeneity across various regions: the central and eastern areas generally showed decline, the northern region remained stable with slight increases, the western region saw significant growth, while the southern area exhibited complex fluctuations. (2) Natural and economic factors had notable impacts on CLUE in Henan province. Among these factors, population and economic factors played a dominant role, whereas average temperature exerted an inhibitory effect on CLUE in most parts of the province. (3) The influenced factors on CLUE varied spatially, with human activity impacts being more concentrated, while topographical and climatic influences were relatively dispersed. These findings provide a scientific basis for land management and agricultural policy formulation in major grain-producing areas, offering valuable insights into enhancing regional CLUE and promoting sustainable agricultural development.

Evaluating Pilot Mental Workload Using fNIRS-Based Functional Connectivity Features with a Deep Residual Shrinkage Network Under Emergency Flight Scenarios

Excessive mental workload can lead to less remaining resources for pilots to perform concurrent tasks during emergency flights, affecting aviation safety. Based on a flight simulator, this study investigated 25 cadet pilots using functional near-infrared spectroscopy (fNIRS) and subjective ratings to assess their mental workload under three subtasks with different equipment failures. fNIRS data included oxyhemoglobin, deoxyhemoglobin, and total hemoglobin signals, yielding 10545 functional connectivity (FC) features from four brain regions: prefrontal, right motor, left motor, and occipital cortexes. A deep residual shrinkage network classified mental workload levels, outperforming convolutional neural network and random forest models with 89.58% accuracy after feature selection employing an interpretable machine learning algorithm. The results suggest that brain FC from three hemoglobin signals could be used to differentiate the three different levels of pilot mental workload. This study could contribute to improving pilot training and supporting the development of pilots’ competencies during emergency scenarios.

Investigation of groundwater quality indices and health risk assessment of water resources of Jiroft city, Iran, by machine learning algorithms

Assessing water quality is essential for acquiring a better understanding of the importance of water in human society. In this study, the quality of groundwater resources in Jiroft city, Iran, using artificial intelligence methods to estimate the groundwater quality index (GWQI) was evaluated. The analysis of hydrochemical parameters, including arsenic (As), fluoride (F), nitrate (NO3), and nitrite (NO2), in 408 samples revealed that concentrations of F, NO3, and NO2 were below the WHO standard threshold, but levels of As exceeded the permissible value. The random forest model with the highest accuracy (R2 = 0.986) was the best prediction model, while logistic regression (R2 = 0.98), decision tree (R2 = 0.979), K-nearest neighbor (R2 = 0.968), artificial neural network (R2 = 0.955), and support vector machine (R2 = 0.928) predicted GWQI with lower accuracy. The non-carcinogenic risk assessment revealed that children had the highest hazard quotient for oral and dermal intake, with values ranging from 0.47 to 13.53 for oral intake and 0.001 to 0.05 for dermal intake. The excess lifetime cancer risk of arsenic for children, adult females, and males was found to be from 2.5 × 10–4 to 7.2 × 10–3, 1.2 × 10–4 to 3.6 × 10–3, and 4.3 × 10–5 to 1.2 × 10–3, respectively. This study suggests that any effort to reduce the arsenic levels in the Jiroft population should take into account the health hazards associated with exposure to arsenic through drinking water.

Root microbiome of Panax ginseng in comparison with three other medicinal plants in the families of Araliaceae and Apiaceae

The intricate interplay between endophytic microorganisms and plants in the upkeep of biodiversity, the stability of communities, and the operation of ecosystems needs to be more adequately extensive. Although root-associated microbial communities of plants have been revealed for the last decade, the understanding of bacterial and fungal communities associated with the roots of medicinal plants remains elusive. To highlight the importance of Panax ginseng Meyer (PG) in our research, we investigated the root endophytic bacterial and fungal communities of Panax ginseng Meyer (PG), alongside Aralia cordata (AC), Angelica gigas (AG), and Peucedanum japonicum (PJ), utilizing amplicon-based community profiling and advanced bioinformatic methodologies. The study aimed to investigate the root-endophytic microbiota of ginseng and three other medicinal plants and identify similarities in microbiome composition across different plant species and families. Results revealed that root-endophytic bacterial and fungal communities were influenced by plant species and phylogenetic differences at the family level. Differential abundance tests and random forest models showed microbial features within the same plant family. PG had a distinctive microbial profile with significant B1653_o_Enterobacterales and F8_o_Helotiales. PG had a core microbiome, B10_Allorhizobium-Neorhizobium-Pararhizobium-Rhizobium, and a more evenly distributed microbial network compared to AG, PJ, and AC. Our research reveals the intricate endophytic microbial communities within the roots of medicinal plants, pinpointing specific taxa that may be pivotal to their medicinal qualities and overall plant health. These insights carry notable implications for future studies, particularly those focused on the endophytes of PG and their secondary metabolites, as they deepen our understanding of plant-microbe interactions and their role in enhancing the plants’ therapeutic potential.

Integrative multi-omics approach using random forest and artificial neural network models for early diagnosis and immune infiltration characterization in ischemic stroke

BackgroundIschemic stroke (IS) is a significant global health issue, causing high rates of morbidity, mortality, and disability. Since conventional Diagnosis methods for IS have several shortcomings. It is critical to create new Diagnosis models in order to enhance existing Diagnosis approaches.MethodsWe utilized gene expression data from the Gene Expression Omnibus (GEO) databases GSE16561 and GSE22255 to identify differentially expressed genes (DEGs) associated with IS. DEGs analysis using the Limma package, as well as GO and KEGG enrichment analyses, were performed. Furthermore, PPI networks were constructed using DEGs from the String database, and Random Forest models were utilized to screen key DEGs. Additionally, an artificial neural network model was developed for IS classification. Use the GSE58294 dataset to evaluate the effectiveness of the scoring model on healthy controls and ischemic stroke samples. The effectiveness of the scoring model was evaluated through AUC analysis, and CIBERSORT analysis was conducted to estimate the immune landscape and explore the correlation between gene expression and immune cell infiltration.ResultsA total of 26 significant DEGs associated with IS were identified. Metascape analysis revealed enriched biological processes and pathways related to IS. 10 key DEGs (ARG1, DUSP1, F13A1, NFIL3, CCR7, ADM, PTGS2, ID3, FAIM3, HLA-DQB1) were selected using Random Forest and artificial neural network models. The area under the ROC curve (AUC) for the IS classification model was found to be near 1, indicating its high accuracy. Additionally, the analysis of the immune landscape demonstrated elevated immune-related networks in IS patients compared to healthy controls.ConclusionThe study uncovers the involvement of specific genes and immune cells in the pathogenesis of IS, suggesting their importance in understanding and potentially targeting the disease.

A Novel Workflow for Mapping Forest Canopy Height by Synergizing ICESat-2 and Multi-Sensor Data

Precise information on forest canopy height (FCH) is critical for forest carbon stocks estimation and management, but mapping continuous FCH with satellite data at regional scale is still a challenge. By fusing ICESat-2, Sentinel-1/2 images and ancillary data, this study aimed to develop a workflow to obtain an FCH map using a machine learning algorithm over large areas. The vegetation-type map was initially produced by a phenology-based spectral feature selection method. A forest characteristic-based model was then proposed to map spatially continuous FCH after a multivariate quality control. Our results show that the overall accuracy (OA) and average F1 Score (F1) for eight main vegetation types were more than 90% and 89%, respectively, and the vegetation-type map agreed well with the census areas. The forest characteristic-based model demonstrated a greater potential in FCH prediction, with an R-value 60.47% greater than the traditional single model, suggesting that the addition of the multivariate quality control and forest structure characteristics could positively contribute to the prediction of FCH. We generated a 30 m continuous FCH map by the forest characteristic-based model and evaluated the product with about 35 km2 of airborne laser scanning (ALS) validation data (R = 0.73, RMSE = 2.99 m), which were 45.34% more precise than the China FCH, 2019. These findings demonstrate the potential of our proposed workflow for monitoring regional continuous FCH, and will greatly benefit accurate forest resources assessment.

Modeling water balance components of conifer species using the Noah-MP model in an eastern Mediterranean ecosystem

Abstract. Few studies have investigated the performance of land surface models for semiarid Mediterranean forests. This study aims to parameterize and test the performance of the Noah-MP land surface model for an eastern Mediterranean ecosystem. To this end, we calibrated the model for root zone soil moisture and transpiration of two conifer species, Pinus brutia, and Cupressus sempervirens, in a plantation forest on the Mediterranean island of Cyprus. The study area has a long-term average annual rainfall of 315 mm. Observations from 4 sap flow and 48 soil moisture sensors, for the period from December 2020 to June 2022, were used for model parameterization. A local sensitivity analysis found that the surface infiltration (REFKDT), hydraulic conductivity (SATDK), and stomatal resistance (RSMIN) parameters had the highest impacts on the water balance components (soil evaporation, tree transpiration, surface runoff, and drainage). The model performed better during the wetter 9-month validation period (379 mm rain) than during the drier 10-month calibration period (175 mm rain). Average soil moisture in the top 60 cm of the soil profile was reasonably well captured for both species (daily Nash–Sutcliffe efficiency > 0.80 for validation). Among the three soil layers, the second layer (20–40 cm) showed better simulation performance during both periods and for both species. The model exhibited limitations with respect to simulating transpiration, particularly during the drier calibration period. The inclusion of a root distribution function in the model, along with the monitoring of soil moisture below the 60 cm soil depth in the field, could improve the accuracy of model simulations in such water-limited ecosystems.

Development and Experimentation of a Universal Array-Style Grain Flow Sensor for Multiple Crops Based on Random Forest Flow Regression Model

To develop a grain flow sensor for combine auger grain outlets, a combine auger elevator was evaluated as the research object. A multi-point distributed array-style differential grain flow sensor for rice and wheat has been developed and tested on three field crops, rice, wheat, and rapeseed. The open system flow test bench was designed to compare the effects of differential processing in the time and frequency domains, as well as different filtering methods on the pre-processing of the collected raw sensor signals. Moreover, a random forest algorithm-based flow regression model was constructed for rice, wheat, and rapeseed based on the comparison of the flow signals of different grains. A weighted multiple linear regression model was constructed as the control group, and both bench and field tests were conducted. The results show that the sensor designed in this study can meet the needs of on-line grain flow monitoring. Meanwhile, the field monitoring errors for rice, wheat, and rapeseed based on the random forest flow regression model were −6.42~8.23%, −7.21~5.71%, and −4.19~4.78%, respectively, significantly better than the control group. The universal array-style grain flow sensor developed in this study provides significant practical value for the promotion and development of precision agriculture.

Unraveling nonlinear effects of environment features on green view index using multiple data sources and explainable machine learning

Urban greening plays a crucial role in maintaining environmental sustainability and enhancing people’s well-being. However, limited by the shortcomings of traditional methods, studying the heterogeneity and nonlinearity between environmental factors and green view index (GVI) still faces many challenges. To address the concerns of nonlinearity, spatial heterogeneity, and interpretability, an interpretable spatial machine learning framework incorporating the Geographically Weighted Random Forest (GWRF) model and the SHapley Additive exPlanation (Shap) model is proposed in this paper. In this paper, we combine multi-source big data, such as Baidu Street View data and remote sensing images, and utilize semantic segmentation models and geographic data processing techniques to study the global and local interpretation of the Beijing region with GVI as the key indicator. Our research results show that: (1) Within the Sixth Ring Road of Beijing, GVI shows significant spatial clustering phenomenon and positive correlation linkage, and at the same time exhibits significant spatial differences; (2) Among many environmental variables, the increase of green coverage rate has the most significant positive effect on GVI, while the increase of building density shows a strong negative correlation with GVI; (3) The performance of the GWRF model in predicting GVI is excellent and far exceeds that of comparison models.; (4) Whether it is the green coverage rate, urban built environment or socioeconomic factors, their influence on GVI shows non-linear characteristics and a certain threshold effect. With the help of these non-linear influences and explicit threshold effects, quantitative analyses of greening are provided, which can help to assist urban planners in making more scientific and rational decisions when allocating greening resources.

Derivation and Validation of Prediction of Preterm preeclampsia using machine learning algorithms.

To develop machine learning models for predicting preterm preeclampsia using the information available before 23 weeks' gestation. This was a secondary analysis of the Nulliparous Pregnancy Outcomes Study: Monitoring Mothers-to-Be cohort. We considered 131 features available before 23 weeks including maternal demographics, obstetrics and family history, social determinants of health, physical activity, nutrition, and early second-trimester ultrasound. Our primary outcome was preterm preeclampsia before 37 weeks. The dataset was randomly split into a training set (70%) and a validation set (30%). Machine learning models using glmnet, multilayer perceptron, random forest, XGBoost, and LightGBM models were developed. Using the machine learning approach that achieved the best area under the curves (AUC), we developed the final model. Further feature selection was conducted from the top 25 important features based on SHapley Additive exPlanations values. The performance of the final model was assessed using the validation dataset. Of 9,467 individuals, 219 (2.3%) had preterm preeclampsia. The AUC of the XGBoost model was the highest (AUC 0.749 [95% confidence interval 0.736-0.762]) compared with other models. Therefore, XGBoost was used to develop models using fewer variables. The XGBoost model with the 8 features (in order of importance: mean uterine artery pulsatility index in the early second trimester, chronic hypertension, pregestational diabetes, uterine artery notch, systolic and diastolic blood pressure in the first trimester, body mass index, and maternal age) was chosen as the final model as it had an AUC of 0.741 (95% confidence interval 0.730-0.752) which was not inferior to the original model (P=0.58). The final model in the validation dataset had an AUC of 0.779 (95% confidence interval 0.722-0.831). An online application of the final model was developed (https://kawakita.shinyapps.io/Preterm_preeclampsia/). Machine learning algorithms using information available before 23 weeks can accurately predict preterm preeclampsia before 37 weeks.

Posture-invariant myoelectric control with self-calibrating random forests

IntroductionMyoelectric control systems translate different patterns of electromyographic (EMG) signals into the control commands of diverse human-machine interfaces via hand gesture recognition, enabling intuitive control of prosthesis and immersive interactions in the metaverse. The effect of arm position is a confounding factor leading to the variability of EMG characteristics. Developing a model with its characteristics and performance invariant across postures, could largely promote the translation of myoelectric control into real world practice.MethodsHere we propose a self-calibrating random forest (RF) model which can (1) be pre-trained on data from many users, then one-shot calibrated on a new user and (2) self-calibrate in an unsupervised and autonomous way to adapt to varying arm positions.ResultsAnalyses on data from 86 participants (66 for pre-training and 20 in real-time evaluation experiments) demonstrate the high generalisability of the proposed RF architecture to varying arm positions.DiscussionOur work promotes the use of simple, explainable, efficient and parallelisable model for posture-invariant myoelectric control.

Response of Topsoil Organic Carbon in the Forests of Northeast China Under Future Climate Scenarios

Soil organic carbon (SOC) serves as a highly sensitive indicator of climate change and plays a crucial role in terrestrial carbon cycles. Evaluating the impact of regional land use changes on SOC stocks is essential for assessing ecological and environmental effects. In this study, we utilized 157 soil samples and 11 environmental variables—including soil properties, topographic factors, and climatic conditions—to develop boosted regression tree (BRT) and random forest (RF) models to estimate topsoil SOC stocks for the year 2015. We used a 10-fold cross-validation approach, along with four validation metrics, to assess model performance. The BRT model demonstrated superior accuracy, with a higher R2 and Lin’s consistency correlation coefficient and a lower mean absolute error and root mean square error compared to the RF model. The key environmental factors influencing SOC stock variability in the BRT model included mean annual temperature, elevation, mean annual precipitation, the topographic wetness index (TWI), and catchment area. Based on this, we employed the space-for-time substitution approach and BRT model to forecast the spatial distribution of soil organic carbon (SOC) stocks in Northeast China’s forested regions under future climate scenarios for the 2050s and 2090s. Our findings indicate that, compared to the 2015 levels, the forecast indicates that SOC stocks will decrease by 122 Tg carbon and 123 Tg carbon under two different future scenarios, SSP245 and SSP585, respectively, by the 2050s. By the 2090s, these figures are expected to decrease further by 127 Tg C and 126 Tg C, respectively. Throughout both future periods, SOC stocks will predominantly be concentrated in the northwest region. This research highlights the necessity of thoroughly considering climatic factors in future studies of regional SOC stock dynamics. Moreover, the high-resolution maps produced in this study offer a scientific foundation for enhancing the implementation of ecological management practices in the forested regions of Northeast China, fostering environmental improvement and bolstering SOC and soil management strategies in response to future climate change.

Unraveling the variability and geoclimatic drivers of daily precipitation concentration over a Western Himalayan Province of India.

A sound understanding of daily precipitation concentration provides critical insights about extreme precipitation and associated hazards such as rainstorms, flash floods, cloud bursts, and soil erosion. Therefore, this study examines the spatiotemporal variability and geoclimatic drivers of daily precipitation concentration using the concentration index (CI), precipitation concentration degree (PCD), Q Index (QI), and Theil index (TI) over a Western Himalayan Province, namely Himachal Pradesh, India. For this, the daily precipitation data pertaining to 111 grids (0.25° × 0.25° resolution), encompassing entire province, have been downloaded from India Meteorological Department (IMD), Pune, for the period 1951 to 2020. The trends in concentration indices have been detected using Mann-Kendall (MK) and Sen's slope methods, whereas their relationship with geoclimatic drivers have been assessed using Pearson's correlation coefficient (physiographic) and random forest regression model (climatic) at annual and monsoon season time scales. The results have revealed a high correlation between CI-TI (> 0.95), followed by QI-TI (> 0.84), QI-PCD (> 0.80), TI-PCD (> 0.78), QI-CI (> 0.65), and CI-PCD (> 0.60), representing nearly similar precipitation concentration distribution. Among the concentration indices, QI has shown maximum sensitivity to long-term changes followed by TI, CI, and PCD at both time scales. Similarly, physiographic factors have maximum effect on QI, showing its greatest sensitivity to latitude, longitude and elevation. While considering the impact of climatic drivers, Atlantic Meridian Oscillation (AMO) and Sunspot activity have exerted maximum influence on annual variability of CI, PCD, QI, and TI, whereas monsoon variability of these indices has been modulated by Pacific Decadal Oscillation (PDO), especially over northern parts of Himachal Pradesh. These findings will assist soil and water resource managers, disaster executives, agricultural stakeholders, and urban planners in understanding the unexplored aspects of daily precipitation concentration and making informed decisions over the mountainous regions of western Himalaya.

Cybersecurity of smart grids: Comparison of machine learning approaches training for anomaly detection

Objectives. The transformation of modern electric grids into decentralized smart grids presents new challenges in the field of cybersecurity. The purpose of this work is to conduct research and analysis into the effectiveness of different machine-learning methods for identifying anomalies in decentralized smart networks, including cyberattacks and emergency modes, as well as to develop recommendations on the optimal combination of these methods for ensuring effective cybersecurity under conditions of changing electrical loads.Methods. We consider several machine learning methods for identifying anomalies in power systems that simulate network behavior under conditions of cyberattacks and emergency modes. The relative effectiveness of such methods as multifractal analysis using wavelets, the Isolation Forest model, local outlier factor (LOF), k-means clustering, and one-class support vector machine (One-Class SVM), is analyzed.Results. The comparison of machine learning methods reveals the varying effectiveness of anomaly detection methods used to detect cyber threats and deviations in electrical systems. Isolation Forest is best at detecting abrupt changes related to cyberattacks with high accuracy and a minimum of false positives. While LOF can also be effective in detecting cyberattacks, its increased sensitivity to minor deviations increases the number of false positives. K-means and One-Class SVMs are less effective in detecting abrupt anomalies but are useful for general clustering of data and detecting both abrupt and smooth changes, respectively.Conclusions. The obtained research results indicate the advantages of using a combination of machine learning algorithms to ensure the reliable protection of smart networks from cyberattacks taking into account the nature of the electrical load.

Estimation of population affinity using cranial measurements acquired in multidetector computed tomography images of Japanese and Malay individuals.

It is imperative in a forensic investigation to determine the identity of an unidentified corpse, for which a crucial starting point is to establish population affinity as part of the biological profile supplied by the forensic anthropologist. The present study investigates the feasibility of using multidetector computed tomography (MDCT) images to quantify craniometric variation between Japanese and Malay populations relative to the estimation of population affinity in a forensic context. The Japanese and Malay samples comprise MDCT scans of 252 (122 female; 130 male) and 182 (84 female; 98 male) adult individuals, respectively. A total of 18 measurements were acquired, and two machine learning methods (random forest modeling, RFM; support vector machine, SVM) were applied to classify population affinity. The accuracy of the two-way pooled-sex model was 88.0% for RFM and 94.5% for SVM, respectively. The four-way population and sex model produced an overall classification accuracy of 81.3% for RFM and 91.7% for SVM. The sex-specific models of population affinity showed correct rates of classification of more than 90% in both females (90.8% for RFM and 97.6% for SVM) and males (91.2% for RFM and 97.4% for SVM). Our findings clearly indicate that the cranial measurements acquired in MDCT images can be used for the forensic classification of Japanese and Malay individuals and thus serve as a reference for forensic anthropologists attempting to identify unidentified remains.

Long-Term or Short-Term? Prediction of Ship Detention Duration Based on Machine Learning

The prevalence of ship deficiencies continues to be a significant issue. Data from the Tokyo Memorandum of Understanding reveals that ship detentions in 2023 surged by more than 80% compared with the previous year. The significant number of detained ships not only disrupts ships’ daily operations but also strains port resources, leading to increased additional costs. In light of this issue, predicting the duration of ship detention becomes crucial, as accurate predictions can assist port managers in resource allocation and provide shipping companies with critical information for operational planning. This study is the first to predict ship detention duration, specifically distinguishing between long-term and short-term detained ships. Initially, key deficiency types influencing the ship detention duration were identified using an improved entropy weight–grey relational analysis. Subsequently, in consideration of the imbalance in data distribution between long-term and short-term detentions, a random forest model capable of handling imbalanced data was applied to classify these two types. The study found that fire safety, propulsion and auxiliary machinery, and pollution prevention are the three most critical deficiency types impacting detention duration; and the random forest model sampled and processed from the data level possessed the best model performance, achieving prediction accuracies of 0.71, 0.72, and 0.85 for bulk carriers, containers, and oil tankers, respectively. This research offers a comprehensive analysis of ship detention duration, making a significant contribution to both the theoretical understanding and practical applications in the maritime industry. Accurately predicting ship detention duration provides valuable insights for stakeholders, enabling them to anticipate potential detention scenarios and thus supporting shipping companies in effective fleet management while assisting port authorities in the optimal allocation of berth resources.

The implications of climate change for New Zealand’s freshwater fish

Climate change is poised to reshape ecological communities globally by driving species into new environments and altering interactions between species. Conservation efforts should not only address current pressures but also plan for future pressures, such as sensitive species moving into degraded environments or arising problematic trophic interactions. This study sought to assess how climate change may affect the end-of-century distributions of New Zealand’s native and non-native freshwater fish, including consequences for the overlap between trout (a non-native sports fish) and native species vulnerable to trout predation. Random forest modelling was used to predict end-of-century distributions for New Zealand’s freshwater fish based on six hydrologically downscaled global climate models across four representative concentration pathways. Severe climate change impacts could drive nine native fish species to extinction or near-extinction and cause substantial declines in another eight native species. Seven non-natives are also predicted to decline substantially, including a 30-40% reduction in the extent of trout. To avert these potential extinctions, it is crucial to mitigate climate change severity and improve land use impacting freshwater ecosystems.