Grid Search Method Research Articles

GridSearch and Data Splitting for Effectiveness Heart Disease Classification

Cardiovascular disease (CVD) is a major global health issue that affects death rates significantly. This research aims to improve the early detection and diagnosis of cardiovascular illness by utilizing machine learning methods, particularly classification algorithms. According to estimates from the World Health Organization (WHO), cardiovascular disease (CVD) caused 17.9 million deaths globally in 2019, or 32% of all fatalities. The treatment and prognosis of cardiovascular illness are greatly improved by early detection and diagnosis. Classification, in particular, machine learning, has become a prominent tool for solving problems connected to heart disease. The main objective of this project is to assess how well Grid Search and various data-sharing methods classify cardiac disease. SVM, Random Forest Classifier, Logistic Regression, Naïve Bayes, Decision Tree Classifier, KNN, and XGBoost Classifier are just a few machine learning methods. The UCI heart disease dataset, which contains information from 303 heart disease patients and 165 healthy participants, is used for the evaluation. Performance parameters like recall, accuracy, precision, and F1 score are considered to evaluate the algorithms' efficacy. The investigation's expected outcomes are intended to increase doctors' ability to diagnose cardiac disease more accurately. Moreover, these results may aid in creating more complex classification models for diagnosing cardiac conditions.

Forecasting hospital discharges for respiratory conditions in Costa Rica using climate and pollution data.

Respiratory diseases represent one of the most significant economic burdens on healthcare systems worldwide. The variation in the increasing number of cases depends greatly on climatic seasonal effects, socioeconomic factors, and pollution. Therefore, understanding these variations and obtaining precise forecasts allows health authorities to make correct decisions regarding the allocation of limited economic and human resources. We aimed to model and forecast weekly hospitalizations due to respiratory conditions in seven regional hospitals in Costa Rica using four statistical learning techniques (Random Forest, XGboost, Facebook's Prophet forecasting model, and an ensemble method combining the above methods), along with 22 climate change indices and aerosol optical depth as an indicator of pollution. Models were trained using data from 2000 to 2018 and were evaluated using data from 2019 as testing data. During the training period, we set up 2-year sliding windows and a 1-year assessment period, along with the grid search method to optimize hyperparameters for each model. The best model for each region was selected using testing data, based on predictive precision and to prevent overfitting. Prediction intervals were then computed using conformal inference. The relative importance of all climatic variables was computed for the best model, and similar patterns in some of the seven regions were observed based on the selected model. Finally, reliable predictions were obtained for each of the seven regional hospitals.

Enhancing deep learning model performance in air quality classification through probabilistic hyperparameter tuning with tree-structured parzen estimators

The research introduces an innovative approach to enhance deep learning models for air quality classification by integrating tree-structured Parzen estimators (TPE) into the hyperparameter tuning process. It applies this approach to CNN, LSTM, DNN, and DBN models and conducts extensive experiments using an air quality dataset, comparing it with grid search, random search, and genetic algorithm methods. The TPE algorithm consistently outperforms these methods, demonstrating improved classification accuracy and generalization. This approach’s potential extends to enriching water quality classification models, contributing to environmental sustainability and resource management. Bridging deep learning with TPE offers a promising solution for optimized air quality classification, supporting informed environmental preservation efforts.

Prediction on the seismic performance limits of reinforced concrete columns based on machine learning method

The drift ratio or lateral deformation is typically applied as the indicator in order to evaluate the earthquake-induced damage, one of the most important issues is to determine the seismic performance level limits. Therefore, this study presents to predict the seismic performance level limits of RC columns by using the machine learning method. Firstly, a test database of the backbone curves of RC columns was established after collecting 754 specimens under axial and lateral loads. Then the seismic performance level limits of all the collected columns were taken out as the input values of machine learning. The correlations among the geometric, mechanical parameters and the performance limits of RC columns were analyzed based on Pearson correlation analysis and mutual information method. Afterward, regression models of seven machine learning methods were established to predict the performance level limits of RC columns, while the hyperparameters of the machine learning models were optimized by the grid search and cross-validation methods. The generalization ability of the prediction models was verified and evaluated by using mean square error, mean absolute error, maximum error and R square, meanwhile, the accuracy of the applied methods was also analyzed. The seismic performance level limits of RC columns determined by the machine learning method can comprehensively consider the influence of geometric and mechanical parameters of RC columns. Combined with the earthquake-induced deformation of RC columns, the seismic damage of RC columns can be evaluated reasonably, which is of great significance for evaluating the seismic damage of building structures. The discussion on the prediction accuracy among different machine learning algorithms is also beneficial for the deformation prediction of other RC components.

Detection of COVID-19 epidemic outbreak using machine learning.

The coronavirus disease (COVID-19) pandemic has spread rapidly across the world, creating an urgent need for predictive models that can help healthcare providers prepare and respond to outbreaks more quickly and effectively, and ultimately improve patient care. Early detection and warning systems are crucial for preventing and controlling epidemic spread. In this study, we aimed to propose a machine learning-based method to predict the transmission trend of COVID-19 and a new approach to detect the start time of new outbreaks by analyzing epidemiological data. We developed a risk index to measure the change in the transmission trend. We applied machine learning (ML) techniques to predict COVID-19 transmission trends, categorized into three labels: decrease (L0), maintain (L1), and increase (L2). We used Support Vector Machine (SVM), Random Forest (RF), and XGBoost (XGB) as ML models. We employed grid search methods to determine the optimal hyperparameters for these three models. We proposed a new method to detect the start time of new outbreaks based on label 2, which was sustained for at least 14 days (i.e., the duration of maintenance). We compared the performance of different ML models to identify the most accurate approach for outbreak detection. We conducted sensitivity analysis for the duration of maintenance between 7 days and 28 days. ML methods demonstrated high accuracy (over 94%) in estimating the classification of the transmission trends. Our proposed method successfully predicted the start time of new outbreaks, enabling us to detect a total of seven estimated outbreaks, while there were five reported outbreaks between March 2020 and October 2022 in Korea. It means that our method could detect minor outbreaks. Among the ML models, the RF and XGB classifiers exhibited the highest accuracy in outbreak detection. The study highlights the strength of our method in accurately predicting the timing of an outbreak using an interpretable and explainable approach. It could provide a standard for predicting the start time of new outbreaks and detecting future transmission trends. This method can contribute to the development of targeted prevention and control measures and enhance resource management during the pandemic.

Bayesian-optimized interpretable surrogate model for seismic demand prediction of urban highway bridges

Urban highway bridges are crucial for regional economic development and population mobility, but they are vulnerable to seismic hazards. Accurately predicting the seismic demands of bridges is essential. Several commonly-used methods for seismic demand estimates, such as cloud analysis, incremental dynamic analysis, and multiple stripe analysis, are computationally expensive and subject to theoretical assumptions and convergence issues, especially for portfolios of bridges. Machine learning techniques can be effective, but their lack of interpretability hinders understanding of the surrogate model's predictions. To address these challenges, this study develops a Bayesian-optimized interpretable ensemble learning surrogate model for predicting and interpreting the critical seismic demands of urban highway bridges.Thousands of continuous prestressed concrete girder bridges with typical geometric configurations in China are generated, and 120 ground motion records with different frequency, waveform, and duration characteristics are chosen. The seismic demand dataset is established based on nonlinear time-history analyses for urban highway bridges, accounting for different types of uncertainties. The surrogate models are developed based on the XGBoost ensemble learning method, with hyperparameters automatically and efficiently determined by Bayesian optimization technique. Compared to the random search and grid search methods, Bayesian optimization is approximately 8 times and 29 times faster, respectively, and it also achieves the highest prediction accuracy. Meanwhile, the relationship of each hyperparameter in the surrogate model is also revealed. The prediction results of the proposed model are compared with other machine learning models. The SHAP algorithm is subsequently used to explain the model predictions regarding the feature importance, global interpretations, and feature dependency analysis. The study quantifies the contribution of each structural feature and seismic intensity measure to the selected critical seismic demand and identifies the interaction effects between different input variables in the prediction.

Seismic Autocorrelation Analysis of Deep Mars

AbstractThe InSight mission deployed one seismic station on Mars, providing a chance to apply single‐station‐based autocorrelation analysis to investigate Martian subsurface structures. However, recent analysis indicated the low‐frequency autocorrelation signals may originate from quasi‐periodic high‐amplitude instrumental “glitches” rather than the reflection response of deep Mars. In this study, we detected and removed these high‐amplitude glitches in raw seismic data and employed autocorrelation on the clean vertical component waveforms filtered between 0.05 and 0.1 Hz. We observed signals at the expected times for the olivine‐wadsleyite transition and core‐mantle boundary (CMB) as estimated by other methods. This result suggests that the low‐frequency autocorrelation signals are the reflection response from the olivine‐wadsleyite transition in the mantle and the Martian CMB region, rather than a noise phenomena. A grid search method to fit the observed PcP waveform was used to identify a layer intermediate in velocity between the Martian mantle and core at the Martian CMB.

Application of supervised learning algorithms for temperature prediction in nucleate flow boiling

This work investigates the use of supervised learning algorithms to predict temperatures in an experimental test bench, which was initially designed for studying nucleate boiling phenomena with ethylene glycol/water mixtures. The proposed predictive model consists of three stages of machine learning. In the first one, a supervised algorithm block is employed to determine whether the critical heat flux (CHF) will be reached within the test bench limits. This classification relies on input parameters including bulk temperature, tilt angle, pressure, and inlet velocity. Once the CHF condition is established, another machine learning algorithm predicts the specific heat flux at which CHF will occur. Subsequently, based on the classification generated by the first block, the evolution of temperature in response to increases in heat flux is predicted using either the previously estimated heat flux or the physical limits of the experimental facility as the stopping criterion. To accomplish all these predictions, the study compares the performance of various algorithms including artificial neural networks, random forest, support vector machine, AdaBoost, and XGBoost. These algorithms were specifically trained using cross-validation and grid search methods to optimize their effectiveness. Results for the CHF classification purpose demonstrate that the support vector machine algorithm performs the best, achieving an F1-score of 0.872 on the testing dataset, while the boosting methods (AdaBoost and XGBoost) exhibit signs of overfitting. In predicting the CHF value, the artificial neural network achieved the lower nMAE on the testing dataset (6.18%). Finally, the validation of the temperature forecasting models, trained on a dataset composed of 314,476 samples, reveals similar performances across all methods, with R2 values greater than 0.95.

Evaluation Modeling of Electric Bus Interior Sound Quality Based on Two Improved XGBoost Algorithms Using GS and PSO

There is no doubt that traffic noise has become one of the main sources of urban noise, and the electric bus, as an important means of transport frequently used by people in daily life, has a direct impact on the psychological and auditory health of passengers due to its interior noise characteristics. Consequently, studying electric bus sound quality is an important way to improve vehicle performance and comfort. In this paper, eight electric buses were selected and 64 noise samples were measured. Acoustic comfort was taken as an evaluation index, professionals were organized to complete the subjective evaluation tests for all noise samples based on rank score comparison (RSC). And nine psycho-acoustic objective parameters such as loudness, sharpness and roughness were calculated using ArtemiS software to establish the sound quality database of electric buses. Aiming at the practical application requirements of high-precision modeling of acoustic comfort in vehicles, this paper presented two improved extreme gradient boosting (XGBoost) algorithms based on grid search (GS) method and particle swarm optimization (PSO), respectively, with objective parameters and acoustic comfort as input and output variables, and established three regression models of standard XGBoost, GS-XGBoost and PSO-XGBoost through data training. Finally, the calculation results of three indexes of average relative error, square root error and correlation coefficient indicate that the proposed PSO-XGBoost model is significantly better than GS-XGBoost and standard XGBoost, with its prediction accuracy as high as 97.6 %. This model is determined as the evaluation model of interior acoustic comfort for this case, providing a key technical support for future sound quality optimization of electric buses.

Crack classification and quantitative evaluation based on dimensionality reduction optimization model of multifeature weak magnetic signal

In this paper, we propose a classification and quantitative method based on the optimization model of dimensionality reduction for detecting crack defects in superalloy turbine disks using multifeature weak magnetic signals. First, the experiment on magnetic weak nondestructive testing was conducted on the defects, resulting in the acquisition of 112 sets of preset magnetic anomaly eigenvalue data. The dimensionality of the multidimensional weak magnetic signal features was then reduced using the principal component analysis method to construct the weak magnetic feature sample library. On this basis, a support vector machine (SVM) model was established to classify and quantitatively evaluate the defects. Genetic algorithm and grid search method were used to optimize the model. Finally, the validity of the optimized SVM model was verified by adding artificial and natural crack defect specimens outside of the sample library. Results show that compared with the original model, the feature dimension reduction and optimization of the two methods enhance the classification accuracy of superalloy surface defects and the quantitative accuracy of defect length, width, and depth. The genetic algorithm significantly improved the classification accuracy, increasing it by 39.33 %. Specifically, the genetic algorithm improved the quantitative accuracy of length and width by 17.9 % and 28.56 % respectively. On the other hand, the grid search method improved the quantitative accuracy of depth the most, with an improvement of 49.83 %. For specimens with natural cracks, the optimized SVM model still demonstrated good classification and quantitative effectiveness in detecting defects, with an accuracy rate of over 85 %.

Research on surface defect identification of steel balls based on improved K-CV parameter optimization support vector machine

Surface defects generated during the production process of steel balls can lead to bearing failures, which makes it crucial to promptly detect and classify these defects. Defects classify is helpful for analysis and improving the production process. An algorithm that incorporates K-fold cross-validation (K-CV) with improved grid search is proposed to optimize the parameters of SVM, in order to detect surface defects with steel balls. Principal Component Analysis (PCA) was employed to reduce the dimensionality of the effective features data. The K-CV algorithm was employed in conjunction with an improved grid search method to find the optimal parameters “c” and “g.” This approach not only reduced the search time but also diminished the influence of individual samples on the model, thereby enhancing its robustness and ultimately improving the classification accuracy. The model’s performance was evaluated using a confusion matrix, and a comparison was made with three other machine learning models. The experimental results demonstrated the effectiveness of the proposed algorithm in classifying defects on highly reflective metal surfaces such as steel balls. The model achieved a classification accuracy of 97.15%.

Machine learning approach to predicting the macro-mechanical properties of rock from the meso-mechanical parameters

The determination of fundamental rock mechanical properties, uniaxial compression strength (UCS) and elastic modulus (E), constitutes a pivotal facet of rock engineering design. However, deriving these properties directly from standard laboratory tests on rock core samples can be challenging, especially when dealing with deep high-stress rock formations and weak fractured strata. Thus, it is crucial to establish a cost-effective and practical approach for predicting the macro-mechanical properties of rocks in situ. In this study, a machine learning approach was proposed to predict UCS and E by upscaling meso-mechanical parameters at particle scale in low-porosity crystalline rocks. To expand the correlation database of rock meso-macro mechanical properties, the meso-mechanical parameters, including the fracture toughness, tensile strength of the rock crystal interface, and the elastic modulus of rock crystal, were accurately measured, using a newly designed mechanical apparatus and a nanoindentation device. The grain-based models implemented in the combined finite discrete element method (FDEM-GBM) were developed based on these experimental results, and their reliability was validated though standard tests. Subsequently, a database, including 225 groups of data, was established using the numerical method. Five machine learning algorithms were applied to develop prediction models for UCS and E through data training in the database. Excellent performance improvement was achieved through the application of the grid-search method. The results indicate the optimized kernel ridge regression (KRR) and gaussian process regression (GPR) models demonstrated excellent performance with relative average errors of 4.9% and 1.1% in predicting UCS and E, respectively. Finally, the predicted values of UCS and E were compared with the experimental results, validating the feasibility of the optimized KRR and GPR models.

Application of the LPPL model in the identification and measurement of structural bubbles in the Chinese stock market

This study proposes innovative methods for determining the starting point and estimating the parameters of the LPPL model for predicting stock market bubbles. These methods address the challenges in determining the starting point of irregularly staged bubbles and the issues of multiple local optima or no solutions in previous prediction algorithms. The proposed LPPL model, incorporating innovative algorithms, is applied to the Chinese A-share market from 2018 to 2021. The starting points of structural bubbles are determined using the difference and grid search methods, and the sequential quadratic programming (SQP) algorithm is used to estimate the parameters of the LPPL model. The study measures and warns about structural bubbles in the A-share market. The results demonstrate that the LPPL model with the innovative algorithm can accurately measure staged and structural bubbles in the A-share market and provide reasonably accurate warnings of bubble bursts, which have been confirmed by the market multiple times.

Unified Fake News Detection System (UFNDS) Framework

The deliberate spread of misleading or inaccurate material pose as authentic news is known as "fake news." Its increasing prevalence calls for the creation of practical strategies to recognize and counteract its negative effects on people and society. Previous methods of identifying fake news depended on linguistic signals and stylistic components. However, these methods faced limitations in terms of their applicability and accuracy. To overcome these constraints, this study proposes the utilization of an extended stacking ensemble classification algorithm (ES-ECA), a machine learning technique designed specifically for detecting fake news. By employing this innovative approach, we aim to surpass the existing barriers and enhance our ability to combat misinformation. The ensemble classifier outperformed the individual classifiers, with an accuracy of 75.18% and an F1-score of 81.81%. These findings imply that the suggested algorithm can be utilized to lessen the negative effects of fake news on society and is efficient at identifying it. The EHT-DL model leverages a multi-step approach to effectively detect fake news. It begins with preprocessing steps such as text normalization, special character handling, stemming, stop word removal, tokenization, and lemmatization. This ensures the dataset is clean and ready for subsequent processing. Feature extraction is performed using TF-IDF, N-grams, and word embeddings scores to capture semantic information and word importance. After that, the dataset is divided into training and testing sets and the deep learning model Dl4jMlpClassifier is used to classify the data. To tackle the drawbacks of existing techniques, the EHT-DL model incorporates efficient hyperparameter tuning. It uses both Grid Search and Random Search methods to optimize the Dl4jMlpClassifier's hyperparameters. By using this method, the model's accuracy and capacity to distinguish between authentic and fraudulent news are both improved. The effectiveness of the EHT-DL model is shown by the experimental findings. Standard assessment criteria including accuracy, precision, recall, and F1-score are used to assess the model. In terms of accuracy and efficiency, comparisons with current methods demonstrate the superiority of the proposed model in identifying bogus news (83.27% accuracy, 80.62% precision, 71.57% recall, and 75.63% f1-score). To increase classification accuracy and resilience, OE-MDL combines the phases of optimized deep learning (ODL) and optimized machine learning (OML). An optimized Multilayer Perceptron serves as the Meta classifier in the OML phase, on top of base classifiers such as optimized RandomForest, optimized J48, optimized SMO, optimized NaiveBayes, and optimized IBk. The experimental findings show that the OE-MDL algorithm performs better than other methods with the maximum recall (85.18%), accuracy (84.27%), precision (74.17%), and F1-Score (79.29%), providing a practical means of halting the spread of false information. The framework for the Unified Fake News Detection System (UFNDS) is revealed. The final result of the research works and demonstrates how the three main stages—Deep Learning and Optimized Ensemble Machine (OE-MDL), Efficient Hyperparameter-Tuned Deep Learning Model (EHT-DL), and Enhanced Stacking Ensemble Classification Algorithm (ES-ECA)—are cohesively incorporated into the UFNDS framework. We examine the UFNDS framework's architectural design and show how flexible it is to the always changing problems associated with fake news identification.

Utilizing a Two-Stage Taguchi Method and Artificial Neural Network for the Precise Forecasting of Cardiovascular Disease Risk.

The complexity of cardiovascular disease onset emphasizes the vital role of early detection in prevention. This study aims to enhance disease prediction accuracy using personal devices, aligning with point-of-care testing (POCT) objectives. This study introduces a two-stage Taguchi optimization (TSTO) method to boost predictive accuracy in an artificial neural network (ANN) model while minimizing computational costs. In the first stage, optimal hyperparameter levels and trends were identified. The second stage determined the best settings for the ANN model's hyperparameters. In this study, we applied the proposed TSTO method with a personal computer to the Kaggle Cardiovascular Disease dataset. Subsequently, we identified the best setting for the hyperparameters of the ANN model, setting the hidden layer to 4, activation function to tanh, optimizer to SGD, learning rate to 0.25, momentum rate to 0.85, and hidden nodes to 10. This setting led to a state-of-the-art accuracy of 74.14% in predicting the risk of cardiovascular disease. Moreover, the proposed TSTO method significantly reduced the number of experiments by a factor of 40.5 compared to the traditional grid search method. The TSTO method accurately predicts cardiovascular risk and conserves computational resources. It is adaptable for low-power devices, aiding the goal of POCT.

A Novel COVID-19 Diagnosis Approach Utilizing a Comprehensive Set of Diagnostic Information (CSDI)

The aim of the study was to develop a computerized method for distinguishing COVID-19-affected cases from cases of pneumonia. This task continues to be a real challenge in the practice of diagnosing COVID-19 disease. In the study, a new approach was proposed, using a comprehensive set of diagnostic information (CSDI) including, among other things, medical history, demographic data, signs and symptoms of the disease, and laboratory results. These data have the advantage of being much more reliable compared with data based on a single source of information, such as radiological imaging. On this basis, a comprehensive process of building predictive models was carried out, including such steps as data preprocessing, feature selection, training, and evaluation of classification models. During the study, 9 different methods for feature selection were used, while the grid search method and 12 popular classification algorithms were employed to build classification models. The most effective model achieved a classification accuracy (ACC) of 85%, a sensitivity (TPR) equal to 83%, and a specificity (TNR) of 88%. The model was built using the random forest method with 15 features selected using the recursive feature elimination selection method. The results provide an opportunity to build a computer system to assist the physician in the diagnosis of the COVID-19 disease.

Fireworks: A Potential Artificial Source for Imaging Near-Surface Structures

Abstract Seismic waves induced by incident acoustic waves from air disturbances can be used to image near-surface structures. In this article, we analyze seismic waveforms recorded by a dense array on the Xishancun landside in Li County, Sichuan Province, southwest China during the Lunar New Year’s Eve (27 January 2017). A total of eight event clusters have been identified as a result of firework explosions. For each cluster, which comprises dozens of individual events with high similarity, we manually pick arrival times of the first event recorded by the array and locate it with a grid-search method. We then rotate three-component waveforms of all events from the east, north, and vertical coordinate system to the local LQT coordinates (L, positive direction perpendicular to the landslide surface and pointing downwards; Q, positive direction is from the launch location of firework to the station along the landslide surface; T, perpendicular to the plane formed by the L and Q directions, and the selected positive direction of the T axis makes LQT form the left-hand coordinate system), and stack the LQT components for those events with cross-correlation values CC ≥ 0.8 with respect to the first event. Characteristics of the stacked LQT components are also examined. The particle motions at each station are retrograde ellipse in the frequency range of ∼5–50 Hz, suggesting air-coupled Rayleigh waves generated by the firework explosions. Spectrograms of the Rayleigh waves also show clear dispersions, which might be used to image near-surface velocity structures. Although we cannot directly extract the phase velocities due to the limitation of the seismic array, our study shows that the fireworks might provide a low-cost and easy-to-use seismic source for imaging near-surface structures.

Physics guided wavelet convolutional neural network for wind-induced vibration modeling with application to structural dynamic reliability analysis

Deep neural network (NN) has become one of the common choices of surrogate model for reliability analysis of structural dynamic response under complex wind loads. However, the cost of data acquisition is usually prohibitive, making it hard for NN training to accurately replace the complex wind-structure coupled system using the limited data. In this work, a physics guided wavelet convolutional NN (PGwCNN) is regarded as surrogate model of a wind turbine tower (WinTT) and utilized for wind effects prediction. The law of dynamics is regarded as the prior knowledge and encoded in the NN, which can enhance the learning capability of the trained model. Then, the performance of the PGwCNN is demonstrated through the dynamic reliability analysis of a high and flexible WinTT using probability density evolution method (PDEM). What’s more, the L2 regularization and dropout layers are utilized to alleviate the overfitting issues. N-fold cross validation and grid search method are adopted to maximize the use of the limited training datasets. Results demonstrate that PGwCNN is capable to reach a compromise between efficiency and accuracy for predicting wind effects on the WinTT. In addition, the PGwCNN can effectively enhance the calculation efficiency of the reliability analysis.

Machine learning approaches for real-time forecasting of solar still distillate output

Solar stills provide a promising avenue for freshwater production in regions grappling with water scarcity, especially remote locales. However, their efficiency is often constrained by the variable climatic conditions. Conventional prediction methods fall short in consistently forecasting the yield, leaving a significant gap in optimizing solar still operations. Recognizing this, the introduction of machine learning becomes pivotal. With a robust predictive model, operators can avoid inefficiencies, inconsistent outputs, and sub-optimal resource utilization. The primary objective of this research is to determine the most suitable machine learning model tailored for predicting solar still output under specific environmental conditions. This research work assessed various machine learning models, including linear regression, decision trees, random forest, support vector machines, and multilayer perceptron. Evaluation metrics encompassed Mean Absolute Error (MAE), cross-validation, grid search, and randomized search techniques. Our results identified the Decision Tree model, registering a MAE of 5.43 and 5.74 through random and grid search methods, respectively, as the preeminent predictor for our dataset. This machine learning-centric methodology elevates the precision of solar still output predictions and paves the way for enhanced solar still designs and superior optimization of solar energy conversion mechanisms.

A comparison of random survival forest and Cox regression for prediction of mortality in patients with hemorrhagic stroke

ObjectiveTo evaluate RSF and Cox models for mortality prediction of hemorrhagic stroke (HS) patients in intensive care unit (ICU).MethodsIn the training set, the optimal models were selected using five-fold cross-validation and grid search method. In the test set, the bootstrap method was used to validate. The area under the curve(AUC) was used for discrimination, Brier Score (BS) was used for calibration, positive predictive value(PPV), negative predictive value(NPV), and F1 score were combined to compare.ResultsA total of 2,990 HS patients were included. For predicting the 7-day mortality, the mean AUCs for RSF and Cox regression were 0.875 and 0.761, while the mean BS were 0.083 and 0.108. For predicting the 28-day mortality, the mean AUCs for RSF and Cox regression were 0.794 and 0.649, while the mean BS were 0.129 and 0.174. The mean AUCs of RSF and Cox versus conventional scores for predicting patients’ 7-day mortality were 0.875 (RSF), 0.761 (COX), 0.736 (SAPS II), 0.723 (OASIS), 0.632 (SIRS), and 0.596 (SOFA), respectively.ConclusionsRSF provided a better clinical reference than Cox. Creatine, temperature, anion gap and sodium were important variables in both models.