Relative Importance Of Variables Research Articles

A study on the promotion of biofertilizer: an application of the plackett-luce model

The research focuses on understanding the elements that drive Agro-input dealers to promote biofertilizers and highlights the challenges encountered in their adoption. With increasing environmental concerns over chemical fertilizers, biofertilizers are promoted as a sustainable alternative that improves soil fertility through beneficial microorganisms. The research was conducted in Hassan district, Karnataka, where 40 dealers were surveyed using structured questionnaires. The Plackett-Luce model was employed to analyze ranked data and understand the relative importance of variables. Results show that high promotional activities (worthiness=0.80, P=<0.001), profitability (worthiness=0.42, P=<0.001) and Service by Company Staff (worthiness=0.23, P=<0.001) significantly impact dealers' promotion efforts, whereas brand image (worthiness=-0.56, P=0.349) has minimal influence. Constraints such as farmers' negative attitudes (worthiness=2.65, P=<0.001), insufficient market facilities (worthiness=1.41, P=<0.005) and limited dealer knowledge (worthiness=-2.02, P=<0.001) were identified as significant barriers. Promotional strategies like field trials (worthiness=0.93, P=<0.001), farmers' meetings (worthiness=0.83, P=<0.001) and jeep campaigns (worthiness=0.61, P=<0.001) emerged as the most effective methods for engaging farmers. The study highlights the need for improved dealer training, better market infrastructure and supportive government policies to enhance biofertilizer promotion, fostering sustainable agricultural practices.

Circumpolar and Regional Seascape Drivers of Genomic Variation in a Southern Ocean Octopus.

Understanding how ecological, environmental and geographic features influence population genetic patterns provides crucial insights into a species' evolutionary history, as well as their vulnerability or resilience under climate change. In the Southern Ocean, population genetic variation is influenced across multiple spatial scales ranging from circum-Antarctic, which encompasses the entire continent, to regional, with varying levels of geographic separation. However, comprehensive analyses testing the relative importance of different environmental and geographic variables on genomic variation across these scales are generally lacking in the Southern Ocean. Here, we examine genome-wide single nucleotide polymorphisms of the Southern Ocean octopus Pareledone turqueti across the Scotia Sea and the Antarctic continental shelf, at depths between 102 and 1342 m, throughout most of this species' range. The circumpolar distribution of P. turqueti is biogeographically structured with a clear signature of isolation-by-geographical distance, but with long-distance genetic connectivity also detected between East and West Antarctica. Genomic variation of P. turqueti was also associated with bottom water temperature at a circumpolar scale, driven by a genotype-temperature association with the warmer sub-Antarctic Shag Rocks and South Georgia. Within the Scotia Sea, geographic distance, oxygen and fine-scale isolation-by-water depth were apparent drivers of genomic variation at regional scales. Putative positive selection of haemocyanin (oxygen transport protein), calcium ion transport and genes linked to RNA modification, detected within the Scotia Sea, suggest physiological adaptation to the regional sharp temperature gradient (~0-+2°C). Overall, we identified seascape drivers of genomic variation in the Southern Ocean at circumpolar and regional scales in P. turqueti and contextualised the role of environmental adaptations in the Southern Ocean.

Characterizing juvenile blacktip shark (Carcharhinus limbatus) nursery areas in the Galapagos: new methods and understandings.

New approaches to abundance surveying utilizing unoccupied aerial vehicles (UAVs) are proving to be effective tools in marine and terrestrial environments. We explored UAV efficacy for surveys in the Galapagos Marine Reserve (GMR), where relative abundance patterns of juvenile sharks and subsequent classifications of putative nursery areas based on environmental drivers are lacking. The UAV method allowed greater temporal and spatial coverage. We expanded classification of shark nursery habitat through monthly drone surveys and environmental data collection at 14 sites around San Cristobal Island (GMR) from November 2018 to July 2019. In the period of surveying, 56 flights identified 453 juvenile Carcharhinus limbatus (blacktip shark). Classification of nurseries followed three criteria, necessitating higher density, short-term residency, and annual site fidelity in target locations. We developed preliminary generalized linear models to elucidate potential environmental parameters influencing the perceived abundance and habitat preference of juveniles. Four sites were identified as either potential nurseries or nursery areas for the target species based on previous excursions. An averaged model was subsequently created from the models found to best explain deviance patterns (R2 = 0.10-0.44) in perceived shark abundance and habitat preference. Relative variable importance (RVI) values further clarified the parameters most associated with higher juvenile presence. This approach provides a systematic method of abundance surveying while simultaneously beginning the process of defining when and where we expect to find higher abundance through environmental modelling of most influential parameters to perceived abundance in this environment.

Enhancing the selection of heart failure patients for invasive procedures: how machine learning can predict the diagnostic yield of endomyocardial biopsy

Abstract Background Determining the etiology of heart failure (HF) is crucial for appropriate treatment. Endomyocardial biopsy (EMB) is an invasive procedure that complements clinical assessment and imaging in diagnosing various cardiac disorders. The diagnostic value of EBM depends upon the anticipated yield of the procedure. Whether results from non-invasive examinations may predict the diagnostic yield from EMB is unclear. Purpose We aimed to investigate the relative importance of clinical and instrumental features in predicting a diagnostic EMB result, using a machine learning (ML) approach. Methods We retrospectively examined 665 consecutive adult patients with HF symptoms who underwent right EBM as part of their diagnostic work-up. The primary outcome was a positive biopsy result with a clear diagnosis. The ML algorithm consisted of Gradient Boosting Machines (GBM), with computation of relative variable importance using conditional importance. We calculated the conditional variable importance for 53 clinical or instrumental variables. Model performance was evaluated using a receiver operating characteristic curve, which was compared to a logistic regression model using the same variables. Validation was performed using a 10-fold cross-validation model. Results Among the 665 patients (50±14 years, 67% male), the EMB provided a diagnosis in 138 cases (20.8%). Patients attaining diagnosis were more often older, had a shorter disease duration, presented with ventricular arrhythmia or cardiac arrest, had a pacemaker or defibrillator, displayed higher creatinine, NT-proBNP and high-sensitive troponin T levels, and had smaller ventricular volumes with higher ejection fraction at CMR. All patients with a diagnostic EMB showed late gadolinium enhancement (LGE) on cardiac magnetic resonance (CMR) in the right ventricle, the right atrium or in the left atrium. The relative importance of different variables predicting outcomes is presented in Figure 1. The top six included four distinct CMR features and two circulating biomarkers (NT-proBNP and high-sensitive troponin T). LGE in the right ventricle and the presence of left ventricular hypertrophy were stronger predictors compared to the remaining variables. The GBM model performed significantly better than the logistic regression model (AUC GBM 0.98, AUC logistic regression 0.93, DeLong’s test p<0.001) (Figure 2). Conclusion Using a ML algorithm, CMR LGE, left ventricular hypertrophy, and elevated cardiac biomarkers, stand out as strong overall predictors of an endomyocardial biopsy with a high diagnostic yield.Figure 1Figure 2

New South Wales alcohol and other drug service providers' perceptions of the relative importance of client variables for determining treatment need.

This study examines alcohol and other drug (AOD) service providers' perceptions of the most important variables (client complexity and demographic) for determining treatment need and intensity of intervention. Online cross-sectional survey of N = 188 clinicians/service managers working in AOD services across metropolitan and regional/rural New South Wales, Australia. Participants ranked the importance of demographic and family factors, substance use, physical health, mental health, functioning and activities of daily living and youth-specific variables in identifying treatment need (five-point Likert scales). More than 90% of participants ranked 43 out 56 potential variables as 'very important'/'essential' in identifying treatment need. The 10 variables most ranked as 'very important'/'essential' were 'pregnant or breastfeeding' (95.2%), 'suicide/self-harm' (95.2%), 'overdose risk' (94.7%), 'abuse/neglect' (among youth/adolescent populations; 94.1%), 'mental health severity' (93.6%), 'dependent children' (93.1%), 'co-existing mental health concerns' (93.0%), 'hospitalisations due to mental health' (92.5%), 'child protection concerns' (among youth/adolescent populations; 92.2%) and 'disability' (91.5%). The 10 variables most commonly ranked as 'slightly important'/'not at all important' included 'citizenship' (63.3%), 'sex' (59.6%), 'country of birth' (54.8%), 'highest education' (50.0%), 'sexual orientation' (44.1%), 'relationship status' (33.5%), 'gender' (31.4%), 'transport' (28.2%), 'employment' (23.9%) and 'refugee status' (24.0%). Some ratings differed by geographic location (metropolitan vs. regional/rural) and job role (allied health worker, nurse, doctor or manager). This study provides insight into service providers' perceptions of treatment need and intensity associated with a range of client factors. It is a first step towards improvements in routine data collections that are used to inform treatment planning.

Current and future habitat suitability modelling of Bambusa teres outside forest areas in Nepal under climate change scenarios

Bambusa teres Buch.-Ham. ex Munro (syn. Bambusa nutans subsp. capulata Stapleton) is a fast-growing perennial bamboo that has ecological, economic, cultural and climate change mitigation benefits. However, information on its current and future potential distribution outside forest areas across Nepal and the key factors affecting its growth and distribution are little known. We used a total of 298 occurrence points obtained from the National Bamboo Resource Assessment and 23 environmental variables to project the distribution of B. teres throughout its potential range in Nepal. Maximum entropy model (MaxEnt) was utilized for this study. We assessed the performance of the model using a receiver operating characteristic curve and evaluated the relative importance of predictor variables through a Jackknife procedure. The model achieved a high level of performance with an area under the curve value of 0.928. Precipitation of the coldest quarter (bio_19), temperature seasonality (bio_4) and precipitation seasonality (bio_15) were the significant contributing variables for the distribution of B. teres. The most suitable habitat for B. teres, with a suitability index >0.6, covered 9264.6 km2, with large sections in Eastern and Central Nepal. However, under future climate change scenarios, the area of suitable habitat for the species is projected to increase across Nepal. This study serves as a baseline for assessing potential climate change impacts on B. teres and will enable the development of adaptive measures to protect and establish various bamboo populations outside forest areas in Nepal and globally.

Habitat requirements of a critically endangered Ninox hawk‐owl: implications for island‐wide restoration

Small, isolated populations can be especially sensitive to poor habitat conditions due to their limited capacity to adapt or disperse. Management of such populations, with an objective to implement the most effective strategies, is a conservation priority. The Norfolk Island morepork (Ninox novaeseelandiae undulata) is a Critically Endangered hawk‐owl endemic to Norfolk Island with an estimated population of 25 individuals. This species faces compounding threats from habitat degradation and loss, as well as inbreeding depression. It presents an exemplar for the development of habitat management strategies for top‐order predators in small, isolated systems. We paired fine‐scale light detection and ranging‐derived habitat variables with tracking data to quantify habitat suitability and the relative importance of environmental variables. Maximum entropy modeling indicated vegetation patch size and canopy height were important habitat variables, with moreporks more likely to be found in larger patches and canopy heights above 7 m. Morepork presence was not strongly influenced by the type of wooded vegetation; rather, the species appears to be a generalist occupier of forested areas. We demonstrate that suitable habitat during the breeding season is currently constrained, with coverage limited to one‐third (34%) of the island and an estimated carrying capacity of 36 morepork pairs. Restoration of native forest in cleared areas adjacent to existing wooded patches, with a specific goal to increase patch size, was the optimized management strategy to increase in situ carrying capacity. This study has demonstrated how habitat suitability modeling can inform targeted habitat restoration to enhance threatened species recovery.

Using Artificial Intelligence Algorithms to Study the Relative Importance of Macroeconomic Variables on Foreign Trade in Iraq

International trade is considered the central link in the complex system of contemporary international economic relations. It links all countries of the world in a unified economic system whose goal is to address economic problems at the international level through developing productive capacity, expanding employment opportunities, and enhancing the flow of production factors between countries to achieve economic growth. Our study aimed to clarify the considerable impact of some macroeconomic variables (exchange rate, gross domestic product, public spending, foreign direct investment) on foreign trade in Iraq and to determine the degree of relative importance of the macroeconomic variables affecting foreign trade in Iraq. The study followed the descriptive analytical approach by collecting data related to the study for the period from 2003 to 2020 and then analyzing the data obtained through artificial neural networks.

Real Environment or Virtual Environment?: Perception Bias Evaluation of Perceived Security in Urban Parks

Urban green spaces, specifically urban parks, contribute to overall human health and well-being. However, physical features of urban green spaces can affect perceived security of its users, ultimately impacting their social life and mental health. Hence, various studies have evaluated perceived security in urban green spaces in virtual and real environment settings. However, studies have not yet evaluated the difference in the perception of perceived security and other physical environmental variables of urban green spaces across virtual and real environment-based methods. Nevertheless, there is a possibility that perception bias can exist across both methodologies. Since Crime Prevention Through Environmental Design (CPTED) is a well-established framework in the field, this study investigates the difference in perceived security across methods using the virtual and real environment in urban parks, using a CPTED classification. Fifteen urban parks were video graphed during the day and night. The study collected 524 and 930 responses from real and virtual environment settings, respectively. The results show that, regardless of the stimulus type, all CPTED variables are significantly correlated with perceived security; however, surveillance, maintenance, and activity support are strongly correlated with perceived security in virtual rather than real environments. Perceptions of territoriality, access control and target hardening significantly differ across both methods. Regardless of user type, PS levels are higher in a real environment. The number of park users affects perceived security more in virtual environments. However, the relative importance of CPTED variables in predicting perceived security is similar in both methods. Future research scholars can benefit from this study by understanding user perception patterns of urban parks in virtual and real environments. Similarly, the study results will guide them in creating virtual and real spaces catering to users’ perceptions. The study's findings can also support future virtual environment-based projects by accounting for potential perception biases when translating results to real environment settings. Further, landscape architects can leverage this research to enhance the design and development of urban parks. From the methodological perspective, this study is applicable to various perception-based studies that can aim to evaluate virtual and real environment-based differences.

The Use of Artificial Intelligence in Predicting Chemotherapy-Induced Toxicities in Metastatic Colorectal Cancer: A Data-Driven Approach for Personalized Oncology.

Machine learning models learn about general behavior from data by finding the relationships between features. Our purpose was to develop a predictive model to identify and predict which subset of colorectal cancer patients are more likely to experience chemotherapy-induced toxicity and to determine the specific attributes that influence the presence of treatment-related side effects. The predictor was general toxicity, and for the construction of our data training, we selected 95 characteristics that represent the health state of 74 patients prior to their first round of chemotherapy. After the data were processed, Random Forest models were trained to offer an optimal balance between accuracy and interpretability. We constructed a machine learning predictor with an emphasis on assessing the importance of numerical and categorical variables in relation to toxicity. The incorporation of artificial intelligence in personalizing colorectal cancer management by anticipating and overseeing toxicities more effectively illustrates a pivotal shift towards more personalized and precise medical care.

Utility of Body Weight, Urine Color, and Thirst Perception (WUT) in Determining Hydration in Young Adults.

The primary aim of this study was to assess the efficacy of the weight, urine, thirst (WUT) framework in predicting dehydration after a body water manipulation protocol, while concurrently determining the individual and interactive contributions of the model components. The total study sample was 93 participants (female, n = 47), recruited from two institutions. Phase 1 involved collecting daily hydration measures from free-living participants (Study 1, 58 participants for 3 days; Study 2, 35 participants for 7 days). Phase 2 entailed a two-hour passive heating protocol, where participants from Study 2 were randomly assigned to one of three groups that manipulated total body water over 24-hours using passive heating and fluid restriction. During each Phase, participants provided urine samples, underwent body mass measurements, and completed questionnaires pertaining to thirst perception. Morning and 24-hour urine samples were assessed for color, osmolality, and specific gravity. Differences between intervention groups, based on the probability of hydration status, were examined (ANOVA) and ridge regression analysis assessed the relative importance of variables within the WUT model. The study revealed significant differences among the intervention groups for predicted probability of dehydration, as determined by changes in body mass (p = 0.001), urine color (p = 0.044), and thirst perception (p < 0.001). Binomial ridge regression indicated that change in body mass (58%) and thirst perception (26%) were the most influential predictors of dehydration. These data support use of an enhanced version of the WUT model, underscoring the significance of changes in body mass and thirst perception in the assessment of hydration status.

The relative importance of environmental variables at different spatial scales in structuring a stream fish metacommunity

The relative importance of environmental variables at different spatial scales in structuring a stream fish metacommunity

Development of machine learning-based models for predicting risk factors in acute cerebral infarction patients: a clinical retrospective study

ObjectivesThe aim of this study was to develop machine learning-based models for predicting acute cerebral infarction (ACI) in patients.MethodsWe extracted the data of ACI patients and non-ACI patients (as control) from two hospitals. The Lasso algorithm was employed to select the most crucial features associated with ACI. Five machine learning algorithms-based models were trained, which was performed with 10-fold cross-validation. Then, the area under the receiver operating characteristic curve (AUC), accuracy, and F1-score were calculated in the training models. Accordingly, the training models with excellent performance was selected as the final predictive model. The relative importance of variables was analyzed and ranked.ResultsA total of 150 patients were diagnosed with ACI (50.00%), with a higher proportion of males (70.67% vs. 44.00%) compared to the non-ACI patients. The logistic regression model exhibited a good performance in predicting ACI in the training set, as evidenced by its highest AUC, accuracy, sensitivity, and F1-score. Furthermore, feature importance analysis showed that blood glucose, gender, smoking history, serum homocysteine, folic acid, and C-reactive protein were the top six crucial variables of the logistic regression.ConclusionsIn our work, the ACI risk prediction model developed by the logistic regression exhibited excellent performance. This could contribute to the identification of risk variables for ACI patients and enables clinicians timely and effective interventions.

Synthesis and characterization of Fe(III)-doped beta-cyclodextrin-grafted chitosan cryogel beads for adsorption of diclofenac in aqueous solutions: Adsorption experiments and deep-learning modeling

Diclofenac (DCF) is frequently detected in aquatic environments, emphasizing the critical need for its efficient removal globally. Here, we present the synthesis of Fe(III)-doped β-CD-grafted chitosan (Fe/β-CD@CS) cryogel beads designed for adsorbing DCF in aqueous solutions. The beads exhibited an average size of 2.94 ± 0.66 mm and a point of zero charge of 8.03. Adsorption experiments demonstrated that the Langmuir kinetic model provided the most accurate description of the kinetic data, while the Redlich–Peterson isotherm offered the best fit for the equilibrium data. The beads showcased a theoretical maximum adsorption capacity of 712.3 mg/g for DCF, with the adsorption process being identified as exothermic. DCF adsorption on the beads was attributed to hydrogen bonding, metal cation-π interactions, and electrostatic interactions. Reusability tests exhibited that the beads could be regenerated using 0.1 M NaOH. To perform deep learning modeling, adsorption experiments (n = 17), designed utilizing central composite design (CCD), were conducted in duplicate. The CCD framework incorporated input variables such as initial DCF concentration, adsorbent dosage, and solution pH, while the output variable was the DCF removal rate. Utilizing the adsorption data, an artificial neural network (ANN) model was constructed with a topology of 3: 7:10:1, featuring 3 input variables, 7 neurons in the first hidden layer, 10 neurons in the second layer, and 1 output variable. Employing the ANN model data, 3-D response surface plots were generated to elucidate the relationship between input variables and DCF removal rate. Additional adsorption tests were conducted to evaluate the developed ANN model, affirming its reliable predictability for the DCF removal rate. Analysis of the relative importance of the input variables revealed the following order of importance: solution pH (100 %) > adsorbent dosage (75.2 %) > initial DCF concentration (57.7 %).

A new spatiotemporal convolutional neural network model for short-term crash prediction

AbstractPredicting short-term traffic crashes is challenging due to an imbalanced data set characterized by excessive zeros in noncrash counts, random crash occurrences, spatiotemporal correlation in crash counts, and inherent heterogeneity. Existing models struggle to effectively address these distinct characteristics in crash data. This paper proposes a new joint model by combining the time-series generalized regression neural network (TGRNN) model and the binomially weighted convolutional neural network (BWCNN) model. The joint model aims to capture all these characteristics in short-term crash prediction. The model was trained and tested using real-world, highly disaggregated traffic data collected with inductive loop detectors on the M1 motorway in the UK in 2019, along with crash data extracted from the UK National Accident Database for the same year. The short-term is defined as a 30-min interval, providing sufficient time for a traffic control center to implement interventions and mitigate potential hazards. The year was segmented into 30-min intervals, resulting in a highly imbalanced data set with over 99.99% noncrash samples. The joint model was applied to predict the probability of a crash occurrence by updating both the crash and traffic data every 30 min. The findings revealed that 75.3% of crashes and 81.6% of noncrash events were correctly predicted in the southbound direction. In the northbound direction, 78.1% of crashes and 80.2% of noncrash events were accurately captured. Causal analysis and model-based interpretation were used to analyze the relative importance of explanatory variables regarding their contribution to crashes. The results reveal that speed variance and speed are the most influential factors contributing to crash occurrence.

Does E-government curb corruption? The moderating role of national culture: a machine learning approach

PurposeThe study aims to investigate the relationship between E-government and corruption. It also examines the moderator role of national culture through Hofstede’s dimensions on the association between E-government and corruption.Design/methodology/approachIn addition to panel regression techniques, the authors use the random forest method to assess the order of importance of all significant variables in determining corruption. The sample of this study consists of 55 countries during 2008–2020 period.FindingsThe results show that E-government is negatively correlated with corruption. The authors also find that both economic and cultural variables play an important role in determining corruption. However, religion has no impact on corruption. The results can potentially assist regulators and policy-makers when trying to control corruption as they should take into consideration the cultural background of citizens when making rules and procedures that aim at reducing corruption.Originality/valueThe current study uses random forests model, which allows the regression of variables based on the construction of a multitude of decision trees. The main contribution of using this model compared to the other regression models used in prior studies is to extract the relative importance of each significant variable. More precisely, it evaluates the rank of importance for each significant variable that drives corruption rather than merely identifying variables that drive corruption regardless of their relative importance.

In-vitro multiparametric analysis of imipenem resistance in bacteria caused by acquisition of foreign DNA and natural mutations

Several studies have demonstrated that antimicrobial resistance (AMR) can result from natural mutations or the acquisition of foreign antimicrobial resistance genes. Many environmental parameters are known to affect AMR, including trace antibiotic concentrations in water, pH, hydraulic retention time, and nutrient concentration, among others. However, most analyses that investigate these parameters analyze them independently. This study investigated the development of natural resistance in Salmonella sp., and the horizontal transfer of resistance genes from an imipenem- resistant (IR) E. coli strain under varying environmental conditions. An automated continuous-culture device was employed to investigate the effect of external parameters of relevance to wastewater treatment plants and to conduct a multiparametric analysis to determine the relative importance of the different variables. In terms of mutation-driven resistance, hydraulic retention time, trace antibiotic concentration, and initial cell concentrations are the leading factors contributing to higher IR in Salmonella spp. In the case of acquired resistance, the initial trace antibiotic concentrations, and hydraulic retention times were also important, but pH, initial nutrient concentration, and, very importantly, the initial concentration of IR E. coli, were the leading factors contributing to IR in Salmonella spp. These results point out some important factors that need to be considered during the design and operation of wastewater treatment plants to minimize the appearance and transmission of antibiotic resistance.

One Man’s Bubble Is Another Man’s Rational Behavior: Comparing Alternative Macroeconomic Hypotheses for the US Housing Market

Competing macroeconomic hypotheses have been developed to explain the US housing market and possible bubble behavior. We employ both seasonally adjusted (SA) and non-seasonally adjusted (NSA) monthly data for about 30 independent variables to examine alternative macro hypotheses for home prices. Using a neural network model as an atheoretical non-linear approach to capture the relative importance of alternative macro variables, we show that these hypotheses generate different macro relevance. As an alternative to testing housing time series, we focus on bubble identification being hypothesis dependent. Model forecast errors (residuals) identify the potential presence of bubbles through standardized residual CUSUM tests for structural breaks. By testing for housing bubbles from these unstructured models, we generate conclusions on the presence of bubbles prior to the Great Financial Crisis and the post-pandemic periods. Competing macro hypotheses or narratives will generate different conclusions on the presence of bubbles and create bubble identification issues.

Deep learning-based quantitative analyses of feedback in the land-atmosphere interactions over the Vietnamese Mekong Delta

During the past several decades, the Vietnamese Mekong Delta (VMD) has experienced many severe droughts, resulting in significant impacts on both agriculture and aquaculture. In the evolution and intensification of droughts, local feedbacks in the Land-Atmosphere (LA) interactions were considered to play a crucial role. It is critical to quantify the impact of LA variables on drought processes and severity with the feedback loop of water and energy balances (e.g., soil moisture-latent and sensible heat-precipitation). In this study, a deep learning model, named Long- and Short-term Time-series Network (LSTNet), was applied to simulate the LA interactions over the VMD. With the ERA5 data as modelling inputs, the role of each key variable (e.g., soil moisture, sensible and latent heat) in the LA interactions over the period of 2011–2020 was quantified, and the variations of their inter-relationships in the future period (2015–2099) were also investigated based on the CMIP6 data. The LSTNet model has demonstrated that the deep learning algorithm can effectively capture the relative importance of key variables in the LA interactions. We found that it is crucial to evaluate the effects of soil moisture and sensible heat on the LA interactions, particularly in the dry periods when negative anomalies in soil moisture and sensible heat would significantly reduce the amount of precipitation. In addition, the decline in soil moisture and the rise in sensible heat are anticipated to further diminish precipitation in the future under the changing climate.

Machine learning-assisted performance prediction from the synthesis conditions of nanofiltration membranes

Optimizing membrane fabrication conditions, including monomer and catalyst concentrations, reaction time, etc., is crucial for achieving high-performance membranes. However, the multitude of variables involved can result in a complex and laborious permutation process, making trial-and-error optimization challenging. To address this issue, we developed a conventional multiple linear regression model and machine learning (ML) models, specifically random forest (RF) and multi-layer perceptron (MLP), to predict the performance of nanofiltration membranes in terms of key parameters such as pure water permeance or PWP (LMH/bar), Na2SO4 rejection (%), and NaCl rejection (%), based on specified input variable ranges. The input variables utilized to construct the models are the concentration of piperazine (PIP) (0.01–2 wt%), the concentration of trimesoyl chloride (TMC) (0.05–0.15 wt%), the concentration of sodium lauryl sulfate (SLS) (0–10 mM), and reaction time (5–1200 s). The conventional regression model failed to predict the membrane performance. The RF model exhibited the best prediction capability for PWP and NaCl rejection with the R2 of 0.9806 and 0.9812 for training, and 0.9669 and 0.9082 for testing, respectively. Similarly, the MLP model outperformed the RF model in predicting Na2SO4 rejection with an R2 of 0.9972 for testing and 0.9844 for training. The best membrane exhibited permeance ranging from 16 to 20 LMH/bar and selectivity between NaCl and Na2SO4 of over 4000, was facilitated by specific conditions. These conditions included lower concentrations of PIP (0.05–0.1 wt%), intermediate concentration of TMC (0.1 wt%), lower concentration of SLS (1 mM), and shorter reaction times (5–30 s). The best-performing ML-based models have been used to provide insights into the relative importance of these input variables in determining membrane performance, thereby aiding in correlating model predictions with fundamental principles of membrane fabrication processes.