Comparison Of Models Research Articles

Predictive value of the random forest model based on bioelectrical impedance analysis parameter trajectories for short-term prognosis in stroke patients

BackgroundThe short-term prognosis of stroke patients is mainly influenced by the severity of the primary disease at admission and the trend of disease development during the acute phase (1–7 days after admission).ObjectiveThe aim of this study is to explore the relationship between the bioelectrical impedance analysis (BIA) parameter trajectories during the acute phase of stroke patients and their short-term prognosis, and to investigate the predictive value of the prediction model constructed using BIA parameter trajectories and clinical indicators at admission for short-term prognosis in stroke patients.MethodsA total of 162 stroke patients were prospectively enrolled, and their clinical indicators at admission and BIA parameters during the first 1–7 days of admission were collected. A Group-Based Trajectory Model (GBTM) was employed to identify different subgroups of longitudinal trajectories of BIA parameters during the first 1–7 days of admission in stroke patients. The random forest algorithm was applied to screen BIA parameter trajectories and clinical indicators with predictive value, construct prediction models, and perform model comparisons. The outcome measure was the Modified Rankin Scale (mRS) score at discharge.ResultsPA in BIA parameters can be divided into four separate trajectory groups. The incidence of poor prognosis (mRS: 4–6) at discharge was significantly higher in the “Low PA Rapid Decline Group” (85.0%) than in the “High PA Stable Group “ (33.3%) and in the “Medium PA Slow Decline Group “(29.5%) (all P < 0.05). In-hospital mortality was the highest in the “Low PA Rapid Decline Group” (60%) compared with the remaining trajectory groups (P < 0.05). Compared with the prediction model with only clinical indicators (Model 1), the prediction model with PA trajectories (Model 2) demonstrated higher predictive accuracy and efficacy. The area under the receiver operating characteristic curve (AUC) of Model 2 was 0.909 [95% CI 0.863, 0.956], integrated discrimination improvement index (IDI), 0.035 (P < 0.001), and net reclassification improvement (NRI), 0.175 (P = 0.031).ConclusionPA trajectories during the first 1–7 days of admission are associated with the short-term prognosis of stroke patients. PA trajectories have additional value in predicting the short-term prognosis of stroke patients.

A 50 year record for perfluoroalkyl acids in the high arctic: implications for global and local transport.

Perfluoroalkyl acids (PFAAs) are persistent compounds that are ubiquitous globally, though some uncertainties remain in the understanding of their long-range transport mechanisms. They are frequently detected in remote locations, where local sources may be unimportant. We collected a 16.5 metre ice core on northern Ellesmere Island, Nunavut, Canada to investigate PFAA deposition trends and transport mechanisms. The dated core represents fifty years of deposition (1967-2016), which accounts for the longest deposition record of perfluoroalkylcarboxylic acids (PFCAs) in the Arctic and the longest record of perfluoroalkylsulfonic acids (PFSAs) globally. PFCAs were detected frequently after the 1990s and have been increasing since. Homologue pair correlations, molar concentration ratios, and model comparisons suggest that PFCAs are primarily formed through oxidation of volatile precursors. PFSAs showed no discernible trend, with concentrations at least an order of magnitude lower than PFCAs. We observed episodic deposition of some PFAAs, notably perfluorooctane sulfonic acid (PFOS) and perfluorobutane sulfonic acid (PFBS) before the 1990s, which may be linked to Arctic military activities. Tracer analysis suggests that marine aerosols and mineral dust are relevant as transport vectors for selected PFAAs during specific time periods. These observations highlight the complex mechanisms responsible for the transport and deposition of PFAAs in the High Arctic.

Forecasting the Risk of Cryptocurrencies: Comparison and Combination of GARCH and Stochastic Volatility Models

Abstract The high returns of cryptocurrencies have attracted many investors in recent years. At the same time the evolution of cryptocurrencies is characterized by extreme volatility. For investors, it is therefore key to gauge the risks related to an investment in cryptocurrencies. We provide a comparison of several GARCH and stochastic volatility models for forecasting the risk of cryptocurrencies over the out-of-sample period from 28.09.2018 to 28.02.2023. It turns out that the widely used GARCH(1,1) does not provide accurate risk predictions. In contrast, adding t-distributed innovations or allowing for regime changes improves the accuracy in both model classes. Finally, we consider a Bayesian decision-guided approach with discount learning to combine the different models and provide robust evidence that combining the model predictions leads to accurate combined risk predictions.

Assessing and mitigating health risks of workers exposed to volatile organic compounds in contaminated soils during active pit excavation: accounting for exposure variability and uncertainty

Health risks to workers involved in the active pit excavation of volatile organic compound (VOC)-contaminated soils have been overlooked, despite the prevalence of ex-situ remediation practices in China. This study developed a new VOC exposure in pit excavation (VEPE) model that comprehensively accounts for the release of gas-phase VOCs in soil pores, surface volatilization of newly exposed agglomerates, and volatilization of newly exposed excavation pit bottom surfaces. Additionally, the model incorporates the concentration variation of soil contaminants, spatial heterogeneity of soil properties, and the variability and uncertainty of worker exposure using two-dimensional Monte Carlo simulation (2D MCS), highlighting the importance of controlling area of new exposure surfaces, soil saturation, and air exchange during the pit excavation to mitigate risks. The study evaluated worker-health-based risks during the excavation of a VOC-contaminated site in Beijing and revealed that the inhalation cancer risk for trichloroethylene (2.35×10-6) and its hazard quotient (143.92) exceed the acceptable levels (10-6 and 1.0, respectively). The comparison of models demonstrated the theoretical advancements of the VEPE model over existing pit excavation models, while the field test comparison showed a significant improvement in the predication accuracy. This study addresses the under-researched area of worker safety in contaminated excavation sites. The findings have the potential to promote cleaner and more sustainable practices in ex-situ contaminated site remediation by providing valuable insights for implementing effective control measures to mitigate the emission of toxic VOCs, thereby reducing adverse social and environmental impacts.

Design of a pharmacokinetic/pharmacodynamic model for administration of low dose peripheral norepinephrine during general anaesthesia.

Intraoperative hypotension is a risk factor for kidney, heart and cognitive postoperative complications. Literature suggests that the use of low-dose peripheral norepinephrine (NOR) reduces organ dysfunction, yet its administration remains unstandardized. In this work we develop a pharmacokinetic (PK)/pharmacodynamic (PD) model of NOR and its effect on mean arterial pressure (MAP). From June 2018 to December 2021, we included patients scheduled for elective neurosurgery and requiring vasopressors for intraoperative hypotension management at Lariboisière Hospital, Paris. Low doses of NOR were administered peripherally, and successive arterial blood samples were collected to track its plasmatic concentration. We used a compartmental modelling approach for NOR PK. We developed and compared 2 models for NOR PD on MAP. Model comparison was done using Bayes information criteria. The resulting PK/PD model parameters were fitted over the entire population and linked to age, weight, height and sex. We included 29 patients (age 52 [46-64] years, 69% female). NOR median time to peak effect on MAP was 74 [53-94] s. After bolus administration, MAP increased by 24% (15-31%). A 2-comparment model with depot best captured NOR PK. NOR PD effect on MAP was well represented by both Emax and Windkessel models, with better results for the former. We found that age, height and weight as well as history of smoking and hypertension were correlated with model parameters. We have developed a PK/PD model to accurately track norepinephrine plasma concentration and its effect on MAP over time, which could serve for target-controlled infusion.

Machine learning for the adsorptive removal of ciprofloxacin using sugarcane bagasse as a low-cost biosorbent: comparison of analytic, mechanistic, and neural network modeling.

Contamination with traces of pharmaceutical compounds, such as ciprofloxacin, has prompted interest in their removal via low-cost, efficient biomass-based adsorption. In this study, classical models, a mechanistic model, and a neural network model were evaluated for predicting ciprofloxacin breakthrough curves in both laboratory- and pilot scales. For the laboratory-scale (d = 2.2cm, Co = 5mg/L, Q = 7mL/min, T = 18°C) and pilot-scale (D = 4.4cm, Co = 5mg/L, Q = 28mL/min, T = 18°C) setups, the experimental adsorption capacities were 2.19 and 2.53mg/g, respectively. The mechanistic model reproduced the breakthrough data with high accuracy on both scales (R2 > 0.4 and X2 < 0.15), and its fit was higher than conventional analytical models, namely the Clark, Modified Dose-Response, and Bohart-Adams models. The neural network model showed the highest level of agreement between predicted and experimental data with values of R2 = 0.993, X2 = 0.0032 (pilot-scale) and R2 = 0.986, X2 = 0.0022 (laboratory-scale). This study demonstrates that machine learning algorithms exhibit great potential for predicting the liquid adsorption of emerging pollutants in fixed bed.

Longitudinal Analysis of Cancer Family Caregiver Perception of Sleep Difficulty During Home Hospice.

Sleep difficulty in caregivers is associated with poor physical and psychological outcomes. The purpose of this study was to describe family caregivers' perception of sleep difficulty through the hospice trajectory after a cancer diagnosis as predicted by age, sex, self-report of anxiety or depression, and cohabitation. We conducted a secondary analysis of longitudinal data using multilevel modeling with nested model comparisons. Beginning with an unconditional growth model, predictors were added to nested models to test differential impact. Caregivers (n = 164) were predominately white (n = 160; 97%) and female (n = 113, 69%). We hypothesized that age, sex, history of anxiety or depression, and cohabitation would predict sleep difficulty. The cohabitation predictor model was a statistically significant model for caregiver perception of sleep difficulty that worsened throughout hospice caregiving (b = .184, 2 = 7.199, P = 0.027) but age, sex, and history of depression or anxiety did not improve model fit. Our findings indicate that family caregivers who cohabitate exhibit increased perception of sleep difficulty over the course of hospice. Future studies and interventions for hospice family caregivers' sleep should consider cohabitation between the patient and the caregiver as a significant predictor of sleep difficulty to observe and potentially mediate the negative outcomes associated with caregiver sleep difficulty. Further, determining the underlying reasons for sleep difficulty in cohabitation (e.g., patient symptoms or treatments) should be explored.

Evaluation and comparison of infiltration models for estimating infiltration capacity of different textures of irrigated soils

Accurate estimation of infiltration rates is crucial for effective irrigation system design and evaluation by optimizing irrigation scheduling, preventing soil erosion, and enhancing water use efficiency. This study evaluates and compares selected infiltration models for estimating water infiltration rates in the Shillanat-iv irrigation scheme in northern Ethiopia. Soil samples were collected to determine textural classes using hydrometer soil texture analysis and the United States Department of Agriculture (USDA) textural triangle. The soil textural map of the study was created using the inverse distance weight interpolation technique in ArcGIS version 10.4. Infiltration rates were measured using the double-ring infiltrometer for five soil textures: clay loam, loam, sandy clay loam, clay, and sandy loam. Six infiltration models (Kostiakov, Modified Kostiakov, Revised Modified Kostiakov, Philip, Horton, and Novel) were employed and evaluated using statistical parameters. Model calibration and validation were conducted using data from 38 points within the study area. The parameter values of the infiltration models were optimized using SPSS statistical software using least-squares errors. The results showed that, Revised Modified Kostiakov, Modified Kostiakov, and Novel infiltration models demonstrated superior capability in estimating infiltration rates for clay loam, loam, and sandy loam soil textures, respectively. Horton's model outperformed other models in estimating infiltration rates for both sandy clay loam and clay soil textures. The appropriately fitted infiltration models can be utilized in designing the irrigation system to estimate the infiltration rate of soil textures within the selected irrigation scheme and at similar sites with comparable soil textures.

Performance Evaluation and Comparison of Deep Neural Network Models for African Soil Properties Prediction

ABSTRACT The significance of soil in agriculture and human survival cannot be overstated. Given its limited supply, improving soil properties is imperative. This study addresses this challenge by predicting five major soil properties—organic carbon, pH values, Mehlich-3 extractable calcium, Mehlich-3 extractable phosphorus, and sand content—utilizing mid-infrared absorbance measurements from the African Soil Information Service (AfSIS) dataset, covering non-desert regions of Africa. With a pressing need for a reliable and efficient model to predict African soil properties using spectral measurements, this study fills a crucial gap, addressing the scarcity of functional soil property databases in Africa. The developed model eliminates costly soil sample preparation and lengthy chemical analysis, applicable in both onsite and laboratory settings for determining soil functional properties. By employing stacked autoencoders for feature dimensionality reduction and combining discrete wavelet transform with two feature selection methods (stepwise regression and random forest) to build robust multi-layer perceptron (MLP) models, the study offers a comprehensive approach. Evaluation metrics including root mean square error (RMSE), scatter index (SI), Variance Accounted For (VAF), Nash-Sutcliffe Model Efficiency (NSE), and Correlation Coefficient (R) demonstrate the superior performance of the MLP model informed by stacked autoencoder-selected features, outperforming models informed by wavelet-transformed features and partial least square regression. This best-performing autoencoder-based model presents a valuable tool for soil scientists tasked with modeling African soil properties.

Comparison of machine learning models in forecasting different ENSO types

Comparison of machine learning models in forecasting different ENSO types

Optimizing Diabetes Prediction with Machine Learning: Model Comparisons and Insights

This study aims to predict diabetes using various machine learning models and compare their performances. The dataset utilized contains health indicators and lifestyle factors from a diverse population. The models evaluated include Random Forest, Logistic Regression, Support Vector Machine (SVM), and Gradient Boosting. Results indicate that Gradient Boosting outperforms other models in terms of accuracy, precision, and recall, making it a robust choice for diabetes prediction. The analysis provides insights into the most significant features contributing to diabetes prediction and highlights the potential of machine learning in medical diagnosis.

Comparison of Dual VET Models in Spain: Analysing Educational Quality from the Perspective of Educational Centres

In Spain, dual vocational training has recently become a fundamental pillar to train the working population. This article focuses on two objectives: to evaluate the quality of two models of dual VET in Spain and to identify their differences in terms of the following dimensions: Teaching Team, Process, Evaluation and Synergies. To achieve these objectives, a quantitative methodology and a cross-sectional design were used to collect data using an ad hoc questionnaire (37 items) developed based on a literature review and the 14 European quality criteria. The sample consisted of 263 educational institutions. Regarding the results for the first objective, Model B implemented EQAVET with higher quality. With regards to the second, researchers observed that all dimensions received a high or very high compliance of 70%. However, discrepancies were primarily found in the training of tutors in training centres, quality certification of dual VET and links, and support and cooperation between training centres and companies. This study provides empirical evidence regarding the factors that lead to higher quality dual VET in schools, based on the manner in which the criteria of the European Framework for Quality and Effective Vocational Education and Training (EQAVET) are implemented.

A machine learning-based model for stroke prediction

In recent years, machine learning has highlighted good results in the early diagnosis and prediction of diseases. Stroke is a serious threat to human health. Early prediction of stroke is of great significance for treatment and intervention. This paper mainly investigates the application of different machine learning models in stroke prediction and compares the performance of each model. First, we collected some data, which contained 5110 entries or records and 12 different attributes. The dataset was then subjected to various preprocessing measures, such as eliminating data redundancy. Seven different machine learning models were used. Includes Logistic Regression, Support Vector Classifier (SVC, K-Neighbors Classifier, Decision Tree Classifier, Random Forest Classifier, XGBoost Classifier (XGBClassifier) and Deep Neural Networks (DNN). After model comparison, we found that when the dataset was extremely imbalanced, the AUC of DNN before feature selection was 82%, which was significantly better than other machine learning models. In addition, the characteristics of each model were analyzed to provide a reference for the selection of stroke prediction models. The results of this study provide value for the early diagnosis and intervention of stroke, and also provide new ideas for the application of machine learning algorithms in the field of stroke prediction.

Biochemical characterization of a SusD-like protein involved in β-1,3-glucan utilization by an uncultured cow rumen Bacteroides.

In ruminants, the rumen is a specialized stomach that is adapted to the breakdown of plant-derived complex polysaccharides through the coordinated activities of a diverse microbial community. Bacteroidota is a major phylum in this bovine rumen microbiota. They contain several clusters of genes called polysaccharide utilization loci (PULs) that encode proteins working in concert to capture, degrade, and transport polysaccharides. Despite the critical role of SusD-like proteins for efficient substrate transport, they remain largely unexplored. Here, we present the biochemical characterization of a SusD-like protein encoded by a β-glucan utilization locus from an Escherichia coli metagenomic clone previously isolated by functional screening of the bovine rumen microbiome. In this study, we show that clone 41O1 can grow on laminaritriose, cellotriose, and a mixture of cellobiosyl-cellobiose and glucosyl-cellotriose as sole carbon sources. Based on this, we used various in vitro analyses to investigate the binding ability of 41O1_SusD-like towards these oligosaccharides and the corresponding polysaccharides. We observed a clear binding affinity for β-1,6 branched β-1,3-glucans (laminarins, yeast β-glucan) and laminaritriose. Comparison of the AlphaFold2 model of 41O1_SusD-like with its closest structural homologs highlights a similar pattern of substrate recognition. In particular, three tryptophan residues are shown to be crucial for laminarin recognition. In the context of the cow rumen, we discuss the possible substrates targeted by the 41O1_PUL, such as the (1,3;1,4)-β-d-glucans present in cereal grains or the β-1,3- and (1,3;1,6)-β-d-glucans that are components of the cell wall of ruminal yeasts.IMPORTANCEThe rumen microbiota can majorly impact overall animal health, feed efficiency, and release of harmful substances into the environment. This microbiota is involved in the fermentation of organic matter to provide the host with valuable and assimilable nutrients. Bacteroidota efficiently captures, breaks down, and imports complex polysaccharides through the concerted action of proteins encoded by polysaccharide utilization loci (PULs). Within this system, SusD-like protein has proven necessary for the active internalization of the substrate. Nevertheless, the vast majority of SusD-like proteins characterized to date originate from cultured bacteria. With regard to the diversity and importance of uncultured bacteria in the rumen, further studies are required to better understand the role of polysaccharide utilization loci in ruminal polysaccharide degradation. Our detailed characterization of the 41O1_SusD-like therefore contributes to a better understanding of the carbohydrate metabolism of an uncultured Bacteroides from the cow rumen.

Calculation of electron-impact ionization of various benzene derivatives.

Abstract Calculations of the electron-impact ionization cross-section of pyrene, anthracene, benzoyl chloride, benzophenone, and phthalonitrile are reported over a wide energy range. A comparison of theoretical models, viz. spherical complex optical potential (SCOP), pixel counting method (PCM), and the binary encounter bethe (BEB) model is carried out. SCOP calculations provide the inelastic cross-section, and ionization cross-section is extracted from it. This result is modified with PCM, a model previously applied to ion collisions which takes geometric screening corrections into account. The BEB model is used as an independent approach to calculate the ionization cross-section. It is demonstrated that all model results are in reasonable agreement with each other. Comparison is also made with other theoretical data where available.&amp;#xD;

Comparison of Process-Driven SWAT Model and Data-Driven Machine Learning Techniques in Simulating Streamflow: A Case Study in the Fenhe River Basin

Hydrological modeling is a crucial tool in hydrology and water resource management for analyzing runoff evolution patterns. In this study, the process-driven soil and water assessment tool (SWAT) model and data-driven machine learning techniques (XGBoost, random forest, LSTM, BILSTM, and GRU) were employed to simulate runoff at monthly and daily intervals in the Fenhe River basin, situated in the middle reaches of the Yellow River, respectively. The SWAT model demonstrated effective performance in simulating runoff at various scales, with the coefficient of determination (R2) exceeding 0.80 and the Nash–Sutcliffe efficiency (NSE) surpassing 0.79. Sensitivity analysis reveals varying degrees of sensitivity among the model parameters. Furthermore, the deep learning techniques (LSTM, BILSTM, and GRU) exhibited superior simulation generalization capabilities compared to the SWAT model across various scales. Additionally, the generalization abilities of traditional machine learning techniques (XGBoost and random forest) were comparable to the SWAT model. This indicates that deep learning techniques demonstrate remarkable stability and generalization capabilities across various scales. This analysis was motivated by the use of external continuous time series data as input and the application of deep learning techniques to internal mechanisms. Moreover, an integrated modeling approach was used to enhance simulation accuracy by combining the SWAT model with machine learning techniques. The results indicate that the integrated modeling approach improves simulation performance across various scales compared to the single-model approach. This research is significant for improving the efficiency of water resource utilization and management in the Fenhe River basin.

Performance Comparison of Empirical Models for Estimating Global Solar Irradiation in the Soudano-Sahelian Zone of Cameroon: The Case of the City of Maroua and Garoua

Performance Comparison of Empirical Models for Estimating Global Solar Irradiation in the Soudano-Sahelian Zone of Cameroon: The Case of the City of Maroua and Garoua

Projection of the prevalence of tracheal, bronchus, and lung cancer incidence using cigarette smoking prevalence in Iran from 1990 to 2018: a comparison of latent period-based models with standard forecasting models

BackgroundSmoking is the major risk factor for tracheal, bronchus, and lung (TBL) cancers. We investigated the feasibility of projecting TBL cancer incidence using smoking incidence rates by incorporating a range of latent periods from the main risk factor exposure to TBL cancer diagnosis.MethodsIn this ecological study, we extracted data on TBL cancer incidence rates in Iran from 1990 to 2018 from the Global Burden of Disease (GBD) database. We also collected data on Iranian cigarette smoking patterns over the past 40 years through a literature review. The weighted average smoking incidence was calculated using a fixed-effects model with Comprehensive Meta-Analysis (CMA) software. Using these data, the five-year TBL cancer incidence in Iran was projected through time series modeling with IT Service Management (ITSM) 2000 software. A second model was developed based on cigarette smoking incidence using linear regression with SPSS (version 22), incorporating different latent periods. The results of these two models were compared to determine the best latent periods.ResultsAn increasing trend in TBL cancer incidence was observed from 2019 to 2023 (first model: 10.30 [95% CI: 9.62, 10.99] to 11.42 [95% CI: 10.85, 11.99] per 100,000 people). In the second model, the most accurate prediction was obtained with latent periods of 17 to 20 years, with the best prediction using a 17-year latent period (10.13 to 11.40 per 100,000 people) and the smallest mean difference of 0.08 (0.84%) per 100,000 people using the standard forecasting model (the ARIMA model).ConclusionProjecting an increase in TBL cancer incidence rates in the future, an optimal latent period of 17 to 20 years between exposure to cigarette smoke and TBL cancer incidence has implications for macrolevel preventive health policymaking to help reduce the burden of TBL cancer in upcoming years.

The advantages behind the efforts of performing higher-order calibration methods – A case study

BackgroundAcross numerous investigations delved into second- and higher-order data, an undoubted finding emerges: models based on such data can effectively exploit the second-order advantage. However, whether further benefits can be achieved by modeling data of higher dimensions remains a subject of inquiry. In this regard, a prevailing question emerges in third-order data-based applications regarding the fundamental need to increase the data dimension and, hence, the data analysis complexity. This study aims to provide meaningful evidence to support the advantages inherent in employing third-order calibration methods despite the associated efforts, such as instrumentation and data analysis complexity. ResultsThis study compares the analytical performance achieved using a third-order calibration method with those obtained from most of the possible second-order calibration approaches derived from the same dataset. This work delves into the structural properties of the data, modeling limitations, and analytical characteristics associated with each model. Additionally, it includes a comprehensive statistical comparison of the models based on their recovery performance. First, the outcomes demonstrate the importance of capitalizing on all available chemical information and harnessing the full potential of data to maximize its benefits. Moreover, the results provide evidence that asserts the fact that third-order calibration methods bring the opportunity to increase the number of analytes that can be simultaneously determined, notwithstanding the need for more tedious experimental protocols, specialized instrumentation (sometimes), and quite complex data analysis. Significancethis research marks the first extensive comparison of third-order data calibration models with possible second-order calibration methods. Moreover, this work pioneers the incorporation of highly challenging non-multilinear data. The advancements detailed in this study emphasize the advantages of third-order data acquisition, notwithstanding the need for more tedious experimental protocols, specialized instrumentation (sometimes), and quite complex data analysis.

Nonlinear modelling of arrays of submerged wave energy converters

Studies of arrays of wave energy converters (WECs) with respect to power absorption and array interactions are often performed using linear models. However, nonlinear effects can be important and may change power estimates and optimal array designs. In this study, we have compared linear predictions of the behaviour of a shallowly submerged, buoyant point absorber with predictions from the nonlinear model SWASH with a WEC incorporated (WEC-SWASH). The latter was first comprehensively validated against 1:20 Froude scaled measured data from physical experiments. WEC-SWASH predictions of body motions and mean power absorption were generally in good agreement with measurements (absolute bias within 25%), although some discrepancies were observed in the body motions, especially when the device exhibited motion instabilities. The validated WEC-SWASH model was then compared with a linear frequency-domain model, as the latter is well-known and widely used because of its computational efficiency. Model comparisons were carried out for both an isolated WEC and small arrays (up to 5 devices). The mean power estimates from the linear model and WEC-SWASH for two representative wave farms showed good agreement for mild waves, with a difference of less than 5%. However, for larger waves, the disagreement increased to about 75 to 85% between the models (for the two wave farms tested). We found that array interactions for these arrays depend on wave amplitude and not just wave frequency. Furthermore, we discovered that the power take-off coefficients optimized using the linear model were not the optimum coefficients for the nonlinear model (WEC-SWASH).