K-fold Cross-validation Research Articles

AntiPhishStack: LSTM-Based Stacked Generalization Model for Optimized Phishing URL Detection

The escalating reliance on revolutionary online web services has introduced heightened security risks, with persistent challenges posed by phishing despite extensive security measures. Traditional phishing systems, reliant on machine learning and manual features, struggle with evolving tactics. Recent advances in deep learning offer promising avenues for tackling novel phishing challenges and malicious URLs. This paper introduces a two-phase stack generalized model named AntiPhishStack, designed to detect phishing sites. The model leverages the learning of URLs and character-level TF-IDF features symmetrically, enhancing its ability to combat emerging phishing threats. In Phase I, features are trained on a base machine learning classifier, employing K-fold cross-validation for robust mean prediction. Phase II employs a two-layered stacked-based LSTM network with five adaptive optimizers for dynamic compilation, ensuring premier prediction on these features. Additionally, the symmetrical predictions from both phases are optimized and integrated to train a meta-XGBoost classifier, contributing to a final robust prediction. The significance of this work lies in advancing phishing detection with AntiPhishStack, operating without prior phishing-specific feature knowledge. Experimental validation on two benchmark datasets, comprising benign and phishing or malicious URLs, demonstrates the model’s exceptional performance, achieving a notable 96.04% accuracy compared to existing studies. This research adds value to the ongoing discourse on symmetry and asymmetry in information security and provides a forward-thinking solution for enhancing network security in the face of evolving cyber threats.

M1A-Ensem: accurate identification of 1-methyladenosine sites through ensemble models

Background1-methyladenosine (m1A) is a variant of methyladenosine that holds a methyl substituent in the 1st position having a prominent role in RNA stability and human metabolites.ObjectiveTraditional approaches, such as mass spectrometry and site-directed mutagenesis, proved to be time-consuming and complicated.MethodologyThe present research focused on the identification of m1A sites within RNA sequences using novel feature development mechanisms. The obtained features were used to train the ensemble models, including blending, boosting, and bagging. Independent testing and k-fold cross validation were then performed on the trained ensemble models.ResultsThe proposed model outperformed the preexisting predictors and revealed optimized scores based on major accuracy metrics.ConclusionFor research purpose, a user-friendly webserver of the proposed model can be accessed through https://taseersuleman-m1a-ensem1.streamlit.app/.

Identification of amnestic mild cognitive impairment by structural and functional MRI using a machine-learning approach

ObjectiveThe importance of early treatment for mild cognitive impairment (MCI) has been extensively shown. However, classifying patients presenting with memory complaints in clinical practice as having MCI vs normal results is difficult. Herein, we assessed the feasibility of applying a machine learning approach based on structural volumes and functional connectomic profiles to classify the cognitive levels of cognitively unimpaired (CU) and amnestic MCI (aMCI) groups. We further applied the same method to distinguish aMCI patients with a single memory impairment from those with multiple memory impairments. MethodsFifty patients with aMCI were enrolled and classified as having either verbal or visual-aMCI (verbal or visual memory impairment), or both aMCI (verbal and visual memory impairments) based on memory test results. In addition, 26 CU patients were enrolled in the control group. All patients underwent structural T1-weighted magnetic resonance imaging (MRI) and resting-state functional MRI. We obtained structural volumes and functional connectomic profiles from structural and functional MRI, respectively, using graph theory. A support vector machine (SVM) algorithm was employed, and k-fold cross-validation was performed to discriminate between groups. ResultsThe SVM classifier based on structural volumes revealed an accuracy of 88.9% at classifying the cognitive levels of patients with CU and aMCI. However, when the structural volumes and functional connectomic profiles were combined, the accuracy increased to 92.9%. In the classification of verbal or visual-aMCI (n = 22) versus both aMCI (n = 28), the SVM classifier based on structural volumes revealed a low accuracy of 36.7%. However, when the structural volumes and functional connectomic profiles were combined, the accuracy increased to 53.1%. ConclusionStructural volumes and functional connectomic profiles obtained using a machine learning approach can be used to classify cognitive levels to distinguish between aMCI and CU patients. In addition, combining the functional connectomic profiles with structural volumes results in a better classification performance than the use of structural volumes alone for identifying both “aMCI versus CU” and “verbal- or visual-aMCI versus both aMCI” patients.

Frequency and Spatial Domain-Based Approaches for Recognition of Indian Sign Language Gestures

Objectives: The objective of this paper is to introduce and demonstrate an innovative approach for the recognition of Indian sign language gestures, with a focus on bridging communication gap between the deaf and hearing communities. The goal is to contribute to the development of effective tools and technologies that facilitate seamless communication between individuals using sign language and the people with no knowledge about sign language. Methods: The methodology consists of three key steps. First, data pre-processing involves resizing and contours extraction. Next, feature extraction employs Fourier descriptors for frequency domain analysis and gray-level-co-occurrence matrix for spatial domain analysis. Finally, various machine learning models including SVM, Random Forest, Logistic Regression, K-Nearest Neighbor and Naive Bayes are trained on a standard dataset. Findings: In our controlled experimental setup, we applied a diverse set of machine learning classifiers to evaluate the proposed approach for gesture recognition. Among the classifiers tested, K-Nearest Neighbors demonstrated the highest accuracy, achieving 99.82%. To validate the robustness of our approach, we employed k-fold cross-validation with 5 folds. Novelty: This study presents an innovative method for sign language recognition by employing a dual-domain fusion strategy that prominently emphasizes the frequency domain. Through the integration of Fourier descriptors, the research conducts a detailed frequency domain analysis to characterize the contour shapes of sign language gestures. The synergy with gray-level co-occurrence matrix texture features in the spatial domain analysis, contributes to the creation of a comprehensive feature vector. The proposed approach ensures a thorough exploration of gesture features, there by advancing the precision and efficacy of sign language recognition. Keywords: Indian Sign Language (ISL), Sign Language Recognition (SLR), Frequency domain, Spatial domain, Fourier descriptors, Gray level co­occurrence matrix (GLCM), K­ Fold

Development of Machine Learning Models for Studying the Premixed Turbulent Combustion of Gas-To-Liquids (GTL) Fuel Blends

AbstractStudying the spatial and temporal evolution in turbulent flames represents one of the most challenging problems in the combustion community. Based on previous 3D numerical analyses, this study aims to develop data-driven machine learning (ML) models for predicting the flame radius evolution and turbulent flame speeds for diesel, gas-to-liquids (GTL), and their 50/50 blend (by volumetric composition) under different thermodynamic and turbulence operating conditions. Two ML models were developed in this study. Model 1 predicts the variations of the flame radius with time, equivalence ratio, and turbulence intensity, whereas model 2 predicts the variations of the turbulence flame speed with the operating parameters. The k-fold cross-validation technique is used for model training, and the developed neural network-based model is used to investigate the effects of operating parameters on the premixed turbulent flames. In addition, the possible minimum and maximum values of responses at the corresponding operating parameters are found using a genetic algorithm (GA) approach. Model 1 could capture the computational fluid dynamics (CFD) outputs with high precision at different flame radiuses and time instants with a maximum absolute error percentage of 5.46%. For model 2, the maximum absolute error percentage was 6.58%. Overall, this study demonstrates the applicability and promising performance of the proposed ML models, which will be used in subsequent research to analyze turbulent flames a posteriori.

Postpartum pharmacologic thromboprophylaxis and complications in a US cohort

Venous thromboembolism accounts for approximately 9% of pregnancy-related deaths in the United States. National guidelines recommend postpartum risk stratification and pharmacologic prophylaxis in at-risk individuals. Knowledge on modern rates of postpartum pharmacologic thromboprophylaxis and its associated risks is limited. This study aimed to describe the rate of, and factors associated with, initiation of postpartum pharmacologic prophylaxis for venous thromboembolism, and to assess associated adverse outcomes. This was a secondary analysis of a multicenter cohort of individuals delivering on randomly selected days at 17 US hospitals (2019-2020). Medical records were reviewed by trained and certified personnel. Those with an antepartum diagnosis of venous thromboembolism, receiving antepartum anticoagulation, or known SARS-CoV-2 infection were excluded. The primary outcome was use of postpartum pharmacologic thromboprophylaxis. Secondary outcomes included bleeding complications, surgical site infection, hospital readmission, and venous thromboembolism through 6 weeks postpartum. The rate of thromboprophylaxis administration was assessed by mode of delivery, institution, and continuance to the outpatient setting. Multivariable regression models were developed using k-fold cross-validation with stepwise backward elimination to evaluate factors associated with thromboprophylaxis administration. Univariable and multivariable logistic models with propensity score covariate adjustment were performed to assess the association between thromboprophylaxis administration and adverse outcomes. Of 21,114 individuals in the analytical cohort, 11.9% (95% confidence interval, 11.4%-12.3%) received postpartum pharmacologic thromboprophylaxis; the frequency of receipt was 29.8% (95% confidence interval, 28.7%-30.9%) following cesarean and 3.5% (95% confidence interval, 3.2%-3.8%) following vaginal delivery. Institutional rates of prophylaxis varied from 0.21% to 34.8%. Most individuals (83.3%) received thromboprophylaxis only as inpatients. In adjusted analysis, cesarean delivery (adjusted odds ratio, 19.17; 95% confidence interval, 16.70-22.00), hysterectomy (adjusted odds ratio, 15.70; 95% confidence interval, 4.35-56.65), and obesity (adjusted odds ratio, 3.45; 95% confidence interval, 3.02-3.95) were the strongest factors associated with thromboprophylaxis administration. Thromboprophylaxis administration was not associated with surgical site infection (0.9% vs 0.6%; odds ratio, 1.48; 95% confidence interval, 0.80-2.74), bleeding complications (0.2% vs 0.1%; odds ratio, 2.60; 95% confidence interval, 0.99-6.80), or postpartum readmission (0.9% vs 0.3%; adjusted odds ratio, 1.38; 95% confidence interval, 0.68-2.81). The overall rate of venous thromboembolism was 0.06% (95% confidence interval, 0.03%-0.10%) and was higher in those receiving prophylaxis (0.2%) compared with those not receiving prophylaxis (0.04%). Approximately 1 in 10 patients received postpartum pharmacologic thromboprophylaxis in this US cohort. Rates of prophylaxis varied widely by institution. Cesarean delivery, hysterectomy, and obesity were predominant factors associated with postpartum thromboprophylaxis administration.

Evaluating the Efficacy of Supervised Machine Learning Models in Inflation Forecasting in Sri Lanka

This study aims to forecast the inflation rate using supervised machine learning models (SMLM). While SMLMs are widely used in various fields, they have not been widely applied in forecasting inflation rates. Therefore, the main objective of this study is to identify the best model for forecasting inflation among four different SMLMs: LASSO regression (LR), Bayesian Ridge Regression (BRR), Support Vector Machine Regression (SVR), and Random Forest Regression (RFR) models. To achieve this objective, two different types of cross-validation techniques were employed: The k-fold cross-validation method (CVK) and walk forward validation (WFV) methods. These techniques were used to estimate the parameters and hyper-parameters for each machine learning model with root mean square error. The mean absolute percentage error (MAPE) was used to compare the performance of the different SMLMs. Empirical evidence from Sri Lanka between 1988 and 2021 was used to test the performance of the SMLMs in forecasting inflation rates. The results show that the SVR model with walk-forward validation is the best method for forecasting the future inflation rate of Sri Lanka based on the MAPE value. Overall, this study showcases the effectiveness of Supervised Machine Learning Models (SMLMs) in forecasting inflation rates, emphasizing the critical role of precise cross-validation techniques. These findings are invaluable for policymakers and investors, offering advanced tools for more informed economic decision-making and highlighting the potential of machine learning in enhancing macroeconomic stability and forecasting accuracy.

Enhancing Energy Efficiency and Classification Modeling Through a Combined Approach of LightGBM and Stratified KFold Cross-Validation

This research article aims to develop a robust and interpretable model for energy efficiency and classification that overcomes the limitations of existing techniques. The study’s objectives are to address the challenges faced by the established methods such as linear regression, random forest, support vector machines, and others. These limitations include issues related to predictive accuracy, model interpretability, and computational efficiency. To bridge these gaps, we introduce a novel model, stratified KFold LightGBM ensemble (SKL-EEFE). SKL-EEFE leverages the strengths of stratified KFold cross-validation, LightGBM gradient boosting, domain-specific feature engineering, and ensemble learning to provide a comprehensive solution to classification tasks. Our key findings reveal that SKL-EEFE outperforms existing methods while maintaining model interpretability. The proposed model offers both accuracy and transparency, making it a valuable addition to the field of classification modeling. This research contributes to advancing the state of knowledge by offering an innovative approach to overcome the limitations of existing models and providing a practical solution for real-world classification problems.

Automated valuation of used sailboat prices based on random forest regression modeling

This study presents a machine learning method for regression prediction of used sailboat prices. The dataset contains attributes such as brand, length, year, and listing price of the sailboat, and the dataset is preprocessed by removing irrelevant fields and normalizing the data. A random forest model is constructed and evaluated against several models such as gradient boosting and neural networks through k-fold cross-validation. Random Forest performs well compared to other models. The ensemble approach of the algorithm effectively modeled the complex nonlinear relationships in the data. Rigorous validation ensures the generalizability of the model. The Random Forest model outperforms traditional manual assessments in terms of the accuracy of price assessments. This data-driven solution allows customers to value sailboats on their own and avoid paying excessive fees. It also allows sailboat companies to develop automated pricing systems to speed up operations. This research provides a powerful machine-learning approach for accurately predicting used sailboat prices. These techniques can be extended to other regression tasks. Further work includes refining the model and deploying real-world applications.

Smartic: A smart tool for Big Data analytics and IoT

The Internet of Things (IoT) is leading the physical and digital world of technology to converge. Real-time and massive scale connections produce a large amount of versatile data, where Big Data comes into the picture. Big Data refers to large, diverse sets of information with dimensions that go beyond the capabilities of widely used database management systems, or standard data processing software tools to manage within a given limit. Almost every big dataset is dirty and may contain missing data, mistyping, inaccuracies, and many more issues that impact Big Data analytics performances. One of the biggest challenges in Big Data analytics is to discover and repair dirty data; failure to do this can lead to inaccurate analytics results and unpredictable conclusions. Different imputation methods were employed in the experimentation with various missing value imputation techniques, and the performances of machine learning (ML) models were compared. A hybrid model that integrates ML and sample-based statistical techniques for missing value imputation is being proposed. Furthermore, the continuation involved the dataset with the best missing value imputation, chosen based on ML model performance for subsequent feature engineering and hyperparameter tuning. K-means clustering and principal component analysis were applied in our study. Accuracy, the evaluated outcome, improved dramatically and proved that the XGBoost model gives very high accuracy at around 0.125 root mean squared logarithmic error (RMSLE). To overcome overfitting, K-fold cross-validation was implemented.

Mission-Driven Inverse Design of Blended Wing Body Aircraft with Machine Learning

The intent of this work was to investigate the feasibility of developing machine learning models for calculating values of airplane configuration design variables when provided time-series, mission-informed performance data. Shallow artificial neural networks were developed, trained, and tested using data pertaining to the blended wing body (BWB) class of aerospace vehicles. Configuration design parameters were varied using a Latin-hypercube sampling scheme. These data were used by a parametric-based BWB configuration generator to create unique BWBs. Performance for each configuration was obtained via a performance estimation tool. Training and testing of neural networks was conducted using a K-fold cross-validation scheme. A random forest approach was used to determine the values of predicted configuration design variables when evaluating neural network accuracy across a blended wing body vehicle survey. The results demonstrated the viability of leveraging neural networks in mission-dependent, inverse design of blended wing bodies. In particular, feed-forward, shallow neural network architectures yielded significantly better predictive accuracy than cascade-forward architectures. Furthermore, for both architectures, increasing the number of neurons in the hidden layer increased the prediction accuracy of configuration design variables by at least 80%.

Imbalance Dataset in Aspect-Based Sentiment Analysis on Game Genshin Impact Review

Sentiment analysis was commonly used to determine the polarity of the review text. However, there is a problem if some reviews have more than one aspect with different polarities, so the reviews have more than one polarity. That has happened in some reviews on the game Genshin Impact. Not merely that, the number of sentiments contained in a review is not always the same as other reviews will cause imbalanced data. So, this study will handle imbalance data with Random Under-Sampling and Random Over-Sampling on aspect-based-sentiment-analysis of Genshin Impact Review with Multinomial Naïve-Bayes, so that the classification prediction does not ignore the minority class due to the dominance of the majority class. The classification process used K-Fold Cross Validation (k=10) validation method and the Laplace smoothing technique on Multinomial Naïve Bayes. As a result, the conclusion is that Random Oversampling had better accuracy than Random Undersampling in handling imbalanced data on aspect-based sentiment analysis of Genshin Impact game Review in Indonesian with Naïve Bayes Multinomial, with the highest accuracy of 85.55%.

Machine-Learning-Based Deformation Prediction Method for Deep Foundation-Pit Enclosure Structure

During the construction of deep foundation pits in subways, it is crucial to closely monitor the horizontal displacement of the pit enclosure to ensure stability and safety, and to reduce the risk of structural damage caused by pit deformations. With advancements in machine-learning (ML) techniques and correlation analysis in engineering, data-driven methods that combine ML with engineering monitoring data have become increasingly popular. These methods offer benefits such as high prediction accuracy, efficiency, and cost effectiveness. The main goal of this study was to develop a machine-learning method for predicting the enclosure deformation of deep foundation pits. This was achieved by analyzing the factors influencing deep foundation-pit enclosure deformation and incorporating historical cases and monitoring reports. The performance of each machine-learning prediction model was systematically analyzed and evaluated using K-Fold cross validation. The results revealed that the random forest model outperformed the other models. The result of the test data showed that the random forest model achieved an R2 of 0.9905, an MAE of 0.8572 mm, and an RMSE of 1.9119 mm. Feature importance analysis identified the depth of enclosure structure, water level, surface settlement, axial force, and exposure time as the most critical factors for accurate prediction. The depth of the enclosure structure had an especially significant impact on the prediction of enclosure deformation.

Sensor-Based Quantification of Peanut Disease Defoliation Using an Unmanned Aircraft System and Multispectral Imagery.

Early leaf spot (Passalora arachidicola) and late leaf spot (Nothopassalora personata) are two of the most economically important foliar fungal diseases of peanut, often requiring seven to eight fungicide applications to protect against defoliation and yield loss. Rust (Puccinia arachidis) may also cause significant defoliation depending on season and location. Sensor technologies are increasingly being utilized to objectively monitor plant disease epidemics for research and supporting integrated management decisions. This study aimed to develop an algorithm to quantify peanut disease defoliation using multispectral imagery captured by an unmanned aircraft system. The algorithm combined the Green Normalized Difference Vegetation Index and the Modified Soil-Adjusted Vegetation Index and included calibration to site-specific peak canopy growth. Beta regression was used to train a model for percent net defoliation with observed visual estimations of the variety 'GA-06G' (0 to 95%) as the target and imagery as the predictor (train: pseudo-R2 = 0.71, test k-fold cross-validation: R2 = 0.84 and RMSE = 4.0%). The model performed well on new data from two field trials not included in model training that compared 25 (R2 = 0.79, RMSE = 3.7%) and seven (R2 = 0.87, RMSE = 9.4%) fungicide programs. This objective method of assessing mid-to-late season disease severity can be used to assist growers with harvest decisions and researchers with reproducible assessment of field experiments. This model will be integrated into future work with proximal ground sensors for pathogen identification and early season disease detection.[Formula: see text] Copyright © 2024 The Author(s). This is an open access article distributed under the CC BY 4.0 International license.

Effect of social capital, social support and social network formation on the quality of life of American adults during COVID-19

This study aims to evaluate the effect of social capital (SC), social support (SS), and social network formation (SNF) on the quality of life of American adults during COVID-19. Using a probability sample of American adults aged 49+, 2370 respondents were selected from the National Social Life Health and Aging Project (NSHAP) dataset for analysis using an integrated partial least squares based on structural equation modeling (PLS-SEM)-K-fold cross-validation approach. The analysis showed that social capital assessed using civic engagement, social cohesion, socioeconomic status (SES), social support, and social network formation were significantly and positively associated with American adults’ quality of life during the COVID-19 pandemic. Furthermore, the results showed that using the PLS-SEM and K-fold cross-validation approach produced a medium predictive power of the overall model, confirming the importance of SC, SS, and SNF in predicting quality of life-outcomes. These findings suggest that efforts to promote the well-being of American adults, especially older adults, during the pandemic should focus on strengthening social capital, social support and social network formation.

Vision-oriented machine learning-assisted seismic energy dissipation estimation for damaged RC beam-column connections

After a seismic event, damage quantification is necessary for safety evaluation, collapse proximity judgment, and measurement of the residual seismic capacity of the joints. Machine learning-based predictive models are developed in this paper aimed at the non-destructive and non-contact estimation of energy dissipation ratio for seismically damaged beam-column joints through surface crack texture complexity indices. For this purpose, the authors collected an experimental dataset comprising 934 images of cracked and/or crushed reinforced concrete-framed joints from 254 specimens under cyclic loading. The associated energy dissipation ratios are then extracted from the lateral load-displacement hysteresis curves for each image by the authors, along with the image-derived fractal indices. Machine learning models with six algorithms, including Linear Regression, Multi-Layer Perceptron, Decision Tree, Random Forest, Gradient Boost, and CatBoost, are trained by 80% of images as a training dataset and validated through the remaining 20% of images in two scenarios. The first scenario develops models by component geometric dimensions and generalized fractal indices calculated for the beams, columns, and joints of seismically damaged connections, while the second scenario adds structural parameters to visual indices to estimate the energy dissipation ratio as the target parameter. Performance evaluation of trained models in the mentioned scenarios via common correlation and accuracy measures, along with the k-fold cross-validation method, indicated that CatBoost-based ML models in both scenarios provide optimal results with R2 = 0.89, RMSE = 0.10 for scenario I and R2 = 0.92, RMSE = 0.09 for scenario II using testing dataset. The first scenario based on the CatBoost algorithm with the R2 variation range of 0.07 and RMSE variation range of 0.03 across the test folds provides a sufficient performance in terms of generalizability and accuracy for estimating the energy dissipation ratio through lower number of input variables.

Metabolic Costs of Walking with Weighted Vests.

The US Army Load Carriage Decision Aid (LCDA) metabolic model is used by militaries across the globe and is intended to predict physiological responses, specifically metabolic costs, in a wide range of dismounted warfighter operations. However, the LCDA has yet to be adapted for vest-borne load carriage, which is commonplace in tactical populations, and differs in energetic costs to backpacking and other forms of load carriage. The purpose of this study is to develop and validate a metabolic model term that accurately estimates the effect of weighted vest loads on standing and walking metabolic rate for military mission-planning and general applications. Twenty healthy, physically active military-age adults (4 women, 16 men; age, 26 ± 8 yr old; height, 1.74 ± 0.09 m; body mass, 81 ± 16 kg) walked for 6 to 21 min with four levels of weighted vest loading (0 to 66% body mass) at up to 11 treadmill speeds (0.45 to 1.97 m·s -1 ). Using indirect calorimetry measurements, we derived a new model term for estimating metabolic rate when carrying vest-borne loads. Model estimates were evaluated internally by k -fold cross-validation and externally against 12 reference datasets (264 total participants). We tested if the 90% confidence interval of the mean paired difference was within equivalence limits equal to 10% of the measured walking metabolic rate. Estimation accuracy, precision, and level of agreement were also evaluated by the bias, standard deviation of paired differences, and concordance correlation coefficient (CCC), respectively. Metabolic rate estimates using the new weighted vest term were statistically equivalent ( P < 0.01) to measured values in the current study (bias, -0.01 ± 0.54 W·kg -1 ; CCC, 0.973) as well as from the 12 reference datasets (bias, -0.16 ± 0.59 W·kg -1 ; CCC, 0.963). The updated LCDA metabolic model calculates accurate predictions of metabolic rate when carrying heavy backpack and vest-borne loads. Tactical populations and recreational athletes that train with weighted vests can confidently use the simplified LCDA metabolic calculator provided as Supplemental Digital Content to estimate metabolic rates for work/rest guidance, training periodization, and nutritional interventions.

A machine learning approach using 18F-FDG PET and enhanced CT scan-based radiomics combined with clinical model to predict pathological complete response in ESCC patients after neoadjuvant chemoradiotherapy and anti-PD-1 inhibitors.

We aim to evaluate the value of an integrated multimodal radiomics with machine learning model to predict the pathological complete response (pCR) of primary tumor in a prospective cohort of esophageal squamous cell carcinoma (ESCC) treated with neoadjuvant chemoradiotherapy (nCRT) and anti-PD-1 inhibitors. Clinical information of 126 ESCC patients were included for analysis. Radiomics features were extracted from 18F-FDG PET and enhanced plan CT images. Four machine learning algorithms, including SVM (Support Vector Machine), Random Forest (RF), and eXtreme Gradient Boosting (XGB) and logistic regression (LR), were applied using k-fold cross-validation to predict pCR after nCRT. The predictive ability of the models was assessed using receiver operating characteristics (ROC) curve analysis. A total of 842 features were extracted. Among the four machine learning algorithms, SVM achieved the most promising performance on the test set for PET(AUC:0.775), CT (AUC:0.710) and clinical model (AUC:0.722). For all combinations of various modalities-based models, the combination model of 18 F-FDG PET, CT and clinical features with SVM machine learning had the highest AUC of 0.852 in the test set when compared to single-modality models in various algorithms. The other combined models had AUC ranged 0.716 to 0.775. Machine learning models utilizing radiomics features from 18F-FDG PET and enhanced plan CT exhibit promising performance in predicting pCR in ESCC after nCRT and anti-PD-1 inhibitors. The fusion of features from multiple modalities radiomics and clinical features enhances the better predictive performance compared to using a single modality alone.

Comparative Analysis Accuracy ID3 Algorithm and C4.5 Algorithm in Selection of Candidates Basic Physics Laboratory Assistant

Basic Physics Laboratory is one of the supporting laboratories at Gunadarma University. Each practical activity in the laboratory is supervised by respective assistants. Therefore, a support system is needed as a basis for decision-making in determining assistant candidates. This decision-making process is processed using data mining techniques, specifically classification algorithms. The criteria or attributes used in the decision-making process include written test scores, practical test scores, presentation scores, equipment usage abilities, and interviews. The classification algorithms used in this research are ID3 and C4.5 algorithms. The tools used to implement these algorithms are RapidMiner Studio 9.10. These algorithms will generate decision trees that can be used as decision support. The aim of this research is to conduct an accuracy comparison analysis for the ID3 and C4.5 algorithms. The highest accuracy obtained will be used as a reference for determining whether assistant candidates are accepted or not. The accuracy results show that the C4.5 algorithm has the highest accuracy, precision, and recall compared to the ID3 algorithm. The determination of the highest value is done using the k-fold cross-validation model for values 2, 4, 6, 8, and 10. The C4.5 algorithm has the highest accuracy of 96.67% at k-fold value = 2.

An open source experimental framework and public dataset for vibration-based fault diagnosis of electrical submersible pumps used on offshore oil exploration

An Electrical Submersible Pump (ESP) is an important equipment used in the industry for lifting liquids in various types of wells. An ESP is widely used in the oil industry for offshore exploration. Detecting a faulty ESP before installation is a predictive maintenance measure in order to extend its operational time. Machine learning fault diagnosis is an effective way for performing this task. Machine learning fault diagnosis algorithms are highly dependent of the availability of an appropriate problem dataset. This paper describes in detail the problem of ESP fault diagnosis and the ESPset dataset, a real-world and public dataset for vibration-based fault diagnosis of electrical submersible pumps used on offshore oil exploration. In addition, the paper also proposes an experimental framework for adequately comparing research works based on the ESPset dataset and defines benchmark classifiers and respective results as referential to the fault diagnosis research community. The framework considers the fact that some subset of samples are not drawn independently, and therefore, proposes a cross-validation sampling strategy that mitigates the similarity bias among samples. Indeed, this work shows that a conventional k-fold cross-validation may lead to a clear overestimation of the average performance. This fact is supported by results which show that the best classification model drops from a mean F-measure of 0.887 to 0.733 when removing the similarity bias from the data.