Search

Lithofacies identification is critical to energy exploration and reservoir evaluation. Machine learning provides a way to use logging data for lithofacies intelligence identification. However, labeled logging data are usually scarce, which makes the currently used supervised algorithms less effective, so semi-supervised methods have received attention from researchers. In this paper, we propose to apply Tri-Training to the field of lithofacies recognition. The framework used Random Forest (RF), Gradient-Boosted Decision Trees (GBDT), and Support Vector Machine (SVM), as the baseline supervised classifiers, and based on the idea of inductive semi-supervised methods and ensemble learning. Baseline classifiers are trained and iterated using unlabeled data to obtain effect improvement. The final results are output in an ensemble paradigm. We used seven logging parameters from two wells as input and divide the data randomly 10 times for training and testing. With only five samples of each lithology, the prediction accuracy improved by the average of 2.1% and 14.5% in both wells compared to the baseline methods. In addition, we also compared two commonly used semi-supervised methods, label propagation algorithm (LPA) and Co-Training. The experimental results also confirm that Tri-training has the better and more stable performance. The Tri-training method in this paper can be effectively applied to lithofacies identification under scarce labeled logging data. [ABSTRACT FROM AUTHOR]

Leptospirosis is a global disease that impacts people worldwide, particularly in humid and tropical regions, and is associated with significant socio-economic deficiencies. Its symptoms are often confused with other syndromes, which can compromise clinical diagnosis and the failure to carry out specific laboratory tests. In this respect, this paper presents a study of three algorithms (Decision Tree, Random Forest and Adaboost) for predicting the outcome (cure or death) of individuals with leptospirosis. Using the records contained in the government National System of Aggressions and Notification (SINAN, in portuguese) from 2007 to 2017, for the state of Pará, Brazil, where the temporal attributes of health care, symptoms (headache, vomiting, jaundice, calf pain) and clinical evolution (renal failure and respiratory changes) were used. In the performance evaluation of the selected models, it was observed that the Random Forest exhibited an accuracy of 90.81% for the training dataset, considering the attributes of experiment 8, and the Decision Tree presented an accuracy of 74.29 for the validation database. So, this result considers the best attributes pointed out by experiment 10: time first symptoms medical attention, time first symptoms ELISA sample collection, medical attention hospital admission time, headache, calf pain, vomiting, jaundice, renal insufficiency, and respiratory alterations. The contribution of this article is the confirmation that artificial intelligence, using the Decision Tree model algorithm, depicting the best choice as the final model to be used in future data for the prediction of human leptospirosis cases, helping in the diagnosis and course of the disease, aiming to avoid the evolution to death. [ABSTRACT FROM AUTHOR]

Intensity modulation/direct detection (IM/DD) optical interconnect systems are susceptible to waveform distortion caused by complex nonlinearities. To reduce the effect of waveform distortion on signal transmission accuracy, in this paper, a random forest (RF) equalization algorithm based on decision tree (DT) integration is proposed for cost-sensitive IM/DD systems, which has the characteristics of nonlinear classification and regression. The performance of the RF equalizer is verified by simulation experiments on a 120 Gbps 8-level pulse amplitude modulation (PAM8) transmission system using an electro-absorption-modulated laser (EML). Simulation studies show that whether it is a back-to-back (B2B) transmission system or a 10 km optical fiber transmission system, the RF equalization scheme can achieve much better bit error rate (BER) performance than the traditional equalization scheme. In addition, in both transmission cases, the RF equalization scheme enables efficient classification of signals and is much less simpler than the artificial neural network (ANN) equalization scheme. [ABSTRACT FROM AUTHOR]

The lithology log, an integral component of the master log, graphically portrays the encountered lithological sequence during drilling operations. In addition to offering real-time cross-sectional insights, lithology logs greatly aid in correlating and evaluating multiple sections efficiently. This paper introduces a novel workflow reliant on an enhanced weighted average ensemble approach for producing high-resolution lithology logs. The research contends with a challenging multiclass imbalanced lithofacies distribution emerging from substantial heterogeneities within subsurface geological structures. Typically, methods to handle imbalanced data, e.g., cost-sensitive learning (CSL), are tailored for issues encountered in binary classification. Error correcting output code (ECOC) originates from decomposition strategies, effectively breaking down multiclass problems into numerous binary subproblems. The database comprises conventional well logs and lithology logs obtained from five proximate wells within a Middle Eastern oilfield. Utilizing well-known machine learning (ML) algorithms, such as support vector machine (SVM), random forest (RF), decision tree (DT), logistic regression (LR), and extreme gradient boosting (XGBoost), as baseline classifiers, this study aims to enhance the accurate prediction of underground lithofacies. Upon recognizing a blind well, the data from the remaining four wells are utilized to train the ML algorithms. After integrating ECOC and CSL techniques with the baseline classifiers, they undergo evaluation. In the initial assessment, both RF and SVM demonstrated superior performance, prompting the development of an enhanced weighted average ensemble based on them. The comprehensive numerical and visual analysis corroborates the outstanding performance of the developed ensemble. The average Kappa statistic of 84.50%, signifying almost-perfect agreement, and mean F-measures of 91.04% emphasize the robustness of the designed ensemble-based workflow during the evaluation of blind well data. [ABSTRACT FROM AUTHOR]

This paper addresses the issue of the prediction of slope stability with machine learning (ML) applications. Five well-known and popular ML algorithms, namely neural network (NNet), decision tree (DT), support vector machine (SVM), k-nearest neighbor (kNN), and random forest (RF), are used to demonstrate the effectiveness of the ML algorithms for predicting binary classification of slope stability based on a case history dataset containing outliers. This study also evaluates the winsorization method used to treat outliers in the dataset by outlining the effect of outliers on the prediction performances of models. To this end, the performance of all the generated ML models is assessed and compared both for unwinsorized (e.g., raw) and winsorized datasets based on performance metrics (i.e., Recall, Precision, Accuracy, and F1-Score) obtained from the confusion matrix. The experimental outputs showed that the application of winsorization enhanced the prediction performance of the models, and thus, all ML models built with winsorized datasets outperformed the unwinsorized ones. In this paper, the RF model achieves the best prediction performance, especially in the case of the winsorized dataset used. Moreover, it is found that SVM is the most sensitive algorithm to outliers as against the other ML algorithms, while the kNN algorithm is the least among the applied algorithms. Results showed that the increment percentage of accuracy nearly reaches 20% for the SVM model and the following 18% for DT, 11% for NNet, 10% for RF, and 4% for kNN, respectively. Furthermore, the results of the study reveal not only the performance of ML algorithms for the slope stability problem but also show how the handling of outliers of a dataset affects the models' prediction performance. [ABSTRACT FROM AUTHOR]

Purpose: Structural damage can significantly alter a system's local flexibility, leading to undesirable displacements and vibrations. Analysing the dynamic structure feature through statistical analysis enables us to discriminate the current structural condition and predict its short- or long-term lifespan. By directly affecting the system's vibration, cracks and discontinuities can be detected, and their severity quantified using the DI. Two damage indexes (DI) are used to build a dataset from the beam's natural frequency and frequency response function (FRF) under both undamaged and damaged conditions, and numerical and experimental tests provided the data-driven. Methods: In this paper, we present the methodology based on machine learning (ML) to monitor the structural integrity of a beam-like structure. The performance of six ML algorithms, including k-nearest neighbors (kNN), Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), and Naive Bayes (NB) are investigated. Results: The paper discusses the challenges of implementing each technique and assesses their performance in accurately classifying the dataset and indicating the beam's integrity. Conclusion: The structural monitoring performed with the ML algorithm achieved excellent metrics when inputting the simulation-generated dataset, up to 100%, and up to 95% having as input dataset provided from experimental tests. Demonstrating that the ML algorithm could correctly classify the health condition of the structure. [ABSTRACT FROM AUTHOR]

Web robots are automated computer programs that can be exploited for benign and malicious activities such as website indexing, monitoring, or unauthorized content scraping and scalping. Several methods are available to detect automated web robots through their footprints and behaviors. Although the accuracy and efficiency of existing methods depend highly on the labeled web log data, countless web requests are generated daily with the help of web robots. Exhaustive and accurate manual labeling of reconstructed sessions is time-consuming and challenging. Further, effective detection of web robots is more challenging with unlabeled or partially labeled data. To address the aforementioned issues, we reformulated web robot detection as a semi-supervised learning problem. In this paper, we propose a deep learning-based Semi-Supervised Stacked Sparse AutoEncoder (Web-S4AE) for web robot detection. The proposed model uses content-based features and features extracted from web access log data to effectively classify web robots. The experiments were conducted on publicly available web log data from a library and information portal to assess the performance of Web-S4AE. The Web-S4AE model was trained in two phases. The first phase; comprises training the model with unlabeled data to extract the hidden information, and in the second phase, the model is fine-tuned using labeled data. The results suggest that incorporating more unlabeled data can significantly improve the classifier's performance. The Web-S4AE model's performance was also compared with other models such as the Decision Tree (DT), Random Forest (RF), eXtreme Gradient Boosting (XGBoost), and Multi-Layer Perceptron (MLP). [ABSTRACT FROM AUTHOR]

Additionally, some of their machine learning model results seem too accurate (it is highly uncommon that a machine learning model would perform better on the test data than the train data, especially for such a small dataset). Machine learning, Scientific knowledge, Data analysis, Model validation Keywords: Machine learning; Scientific knowledge; Data analysis; Model validation EN Machine learning Scientific knowledge Data analysis Model validation 6167 6168 2 07/25/23 20230801 NES 230801 Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s00603-023-03333-z. [Extracted from the article]

Different modeling techniques must be applied to manage production and statistical estimation to predict the expected harvest. By calculating advanced production methods and the rational valuation of different factors, we can accurately capture the variety of growth characteristics and the expected yield. This paper obtained 32 feature variables related to melons, including phenological features, shape features, and color features. The Gradient Boosted Decision Tree (GBDT) network and the Grid Search (GS) hyperparameter seeking method was applied to calculate the degree of importance of all melon fruits' characteristics and construct prediction models for three expected harvest indexes of melon yield, sugar content, and endocarp hardness. To facilitate growers to carry out prediction and estimation in the field without destroying the melon fruits. The reduced feature variables were selected as inputs. The GBDT model was used to provide a significant advantage in prediction compared to both Random Forest (RF) and Support Vector Regression (SVR) methods. In addition, to verify the feasibility of using only reduced feature variables as input for the evaluation work, this study also compares the predictive effects of the model when all feature variables and only reduced feature variables are used. The GBDT prediction model proposed in this paper predicted melon yield, sugar content, and hardness using reduced features as input, and the model R2 could reach more than 90%. Therefore, this method can effectively help growers carry out early non-destructive inspection and growth prediction of melons in the field. [ABSTRACT FROM AUTHOR]

Hypertension is a major risk factor for cardiovascular diseases (CVD). Identifying the persons at these high risks plays an important role because it would save time and money before using any complex, invasive and/or expensive diagnostic methods. This task can be partly dealt with the help of advanced machine-learning methods. Specifically, a prediction model can be developed based on some indicator factors of the people at high risks which are easily-obtained, non-invasive and low-cost. This paper presents the first work towards predicting hypertension risks based on of the people in the Northwest region of Vietnam where hypertension rate is increasing. 2.509 samples were collected and classified into two levels of hypertension or no hypertension. We investigated and compared the performance of robust machine learning methods including single classifiers such as Naïve Bayes, MLP, Decision Tree, kNN and SVM; and ensemble classifiers such as bagging, boosting and voting methods, to generate mathematical models to predict the risk of hypertension disease. The experimental results showed that the best random forest model yielded 72.39% in the F1 score. This result was quite promising and can be applied in Vietnamese hospital nowadays. [ABSTRACT FROM AUTHOR]

This paper compares the classification performance of machine learning classifiers vs. deep learning-based handcrafted models and various pretrained deep networks. The proposed study performs a comprehensive analysis of object classification techniques implemented on low-altitude UAV datasets using various machine and deep learning models. Multiple UAV object classification is performed through widely deployed machine learning-based classifiers such as K nearest neighbor, decision trees, naïve Bayes, random forest, a deep handcrafted model based on convolutional layers, and pretrained deep models. The best result obtained using random forest classifiers on the UAV dataset is 90%. The handcrafted deep model's accuracy score suggests the efficacy of deep models over machine learning-based classifiers in low-altitude aerial images. This model attains 92.48% accuracy, which is a significant improvement over machine learning-based classifiers. Thereafter, we analyze several pretrained deep learning models, such as VGG-D, InceptionV3, DenseNet, Inception-ResNetV4, and Xception. The experimental assessment demonstrates nearly 100% accuracy values using pretrained VGG16- and VGG19-based deep networks. This paper provides a compilation of machine learning-based classifiers and pretrained deep learning models and a comprehensive classification report for the respective performance measures. [ABSTRACT FROM AUTHOR]

The paper explores the possibilities of using data classification methods when forecasting time series of the geomagnetic Kp-index by machine learning methods. To classify categories of the Kp-index based on the degree of disturbance, linear and logistic regression, random forest, gradient boosting on top of decision trees, and artificial neural networks of various architectures are used. The results of these methods are compared with a trivial inertial forecast (the statistical indicators of which for problems of this type are always high) at horizons from 3 h to 1 day in 3-h increments. The problem of choosing a cross-validation scheme for selecting the model hyperparameters, ways to overcome the imbalance of categories, the relative importance of input features, as well as the dependence of the results on the test sample (beginning of the 25th solar activity cycle) on inclusion in the training sample of data from the 23rd and 24th cycles or only the 24th cycles are studied. Based on the results, conclusions are drawn about the preferred methods for classifying values of the Kp-index based on the level of geomagnetic disturbance. Ways for further research and possible improvement of the classification quality are outlined, including for determining the characteristic hidden states of Earth's magnetosphere as a dynamic system in order to improve the quality of forecasting geomagnetic indices. [ABSTRACT FROM AUTHOR]

In recent years, machine learning (ML) has become increasingly popular in various fields of activity. Cloud platforms have also grown in popularity, as they offer services that are more secure and accessible worldwide. In this context, cloud-based technologies emerged to support ML, giving rise to the machine learning as a service (MLaaS) concept. However, the clients accessing ML services in order to obtain classification results on private data may be reluctant to upload sensitive information to cloud. The model owners may also prefer not to outsource their models in order to prevent model inversion attacks and to protect intellectual property. The privacy-preserving evaluation of ML models is possible through multi-key homomorphic encryption (MKHE), that allows both the client data and the model to be encrypted under different keys. In this paper, we propose an MKHE evaluation method for decision trees and we extend the proposed method for random forests. Each decision tree is evaluated as a single lookup table, and voting is performed at the level of groups of decision trees in the random forest. We provide both theoretical and experimental evaluations for the proposed method. The aim is to minimize the performance degradation introduced by the encrypted model compared to a plaintext model while also obtaining practical classification times. In our experiments with the proposed MKHE random forest evaluation method, we obtained minimal (less than 0.6%) impact on the main ML performance metrics considered for each scenario, while also achieving reasonable classification times (of the order of seconds). [ABSTRACT FROM AUTHOR]

Diabetes Mellitus (DM) is a widespread condition that is one of the main causes of health disasters around the world, and health monitoring is one of the sustainable development topics. Currently, the Internet of Things (IoT) and Machine Learning (ML) technologies work together to provide a reliable method of monitoring and predicting Diabetes Mellitus. In this paper, we present the performance of a model for patient real-time data collection that employs the Hybrid Enhanced Adaptive Data Rate (HEADR) algorithm for the Long-Range (LoRa) protocol of the IoT. On the Contiki Cooja simulator, the LoRa protocol's performance is measured in terms of high dissemination and dynamic data transmission range allocation. Furthermore, by employing classification methods for the detection of diabetes severity levels on acquired data via the LoRa (HEADR) protocol, Machine Learning prediction takes place. For prediction, a variety of Machine Learning classifiers are employed, and the final results are compared with the already existing models where the Random Forest and Decision Tree classifiers outperform the others in terms of precision, recall, F-measure, and receiver operating curve (ROC) in the Python programming language. We also discovered that using k-fold cross-validation on k-neighbors, Logistic regression (LR), and Gaussian Nave Bayes (GNB) classifiers boosted the accuracy. [ABSTRACT FROM AUTHOR]

The prediction accuracy of current mainstream machine learning (ML) models depends on regulating many hyperparameters. In this paper, a deep forest (DF) model with a few hyperparameters and a non-excessive dependence on super parameter regulation was applied to the prediction of glass-forming ability (GFA) of bulk metallic glasses (BMGs). Compared with these of the mainstream ML models, including Support Vector Regression (SVR), random forest (RF), gradient boosted decision trees (GBDT), k-nearest neighbor (KNN), and eXtreme gradient boosting (XGBoost), the tenfold cross-validation shows that the determination coefficient (R2) of our suggested DF model is improved by 10.4%–74.2%. Moreover, the parameter Φ obtained by the SHapley Additive exPlanations (SHAP) method analysis can be used to guide the design and development of BMGs. Finally, a design and development of scheme process for BMGs that meets the expected requirements is given via parameter Φ and the constructed DF model. [ABSTRACT FROM AUTHOR]

Formation resistivity is crucial for calculating water saturation, which, in turn, is used to estimate the stock-tank oil initially in place. However, obtaining a complete resistivity log can be challenging due to high costs, equipment failure, or data loss. To overcome this issue, this study introduces novel machine learning models that can be used to predict the electrical resistivity of oil wells, using conventional well logs. The analysis utilized gamma-ray (GR), delta time compressional logs (DTC), sonic shear log (DSTM), neutron porosity, and bulk density. The study utilized a dataset of 3529 logging data points from horizontal oil carbonate wells which were used to develop different machine learning models using random forest (RF) and decision tree (DT) algorithms. The obtained results showed that both models can predict electrical resistivity with high accuracy, over 0.94 for training and testing data. Comparing the models based on accuracy and consistency revealed that the RF model had a slight advantage over the DT model. Based on the data analysis, it was found that the formation resistivity is more significantly impacted by GR logs compared to DTC logs. These new ML models offer a low-cost and practical alternative to estimate well resistivity in oil wells, providing valuable information for geophysical and geological interpretation. [ABSTRACT FROM AUTHOR]

Privacy and security are the most essential barriers preventing widespread public cloud adaptation. The data that is distributed within the cloud requires privacy protection which can be achieved through the implementation of a data anonymization process. However, there is a lack of privacy for the user data. Even though the differential privacy model minimizes the efficiency, it also leads to generate data loss and privacy is not enhanced. Data securing in a cloud environment is a complex task. The data that are distributed within the cloud requires privacy protection which can be achieved through the implementation of a data anonymization process. So in order to improve the efficiency the novel hybrid ABRF-Stacked Bi-GRU is proposed to secure the medical data. This proposed method diminished the leakage of data and also effectively improve security and privacy. Therefore, in this paper, the privacy-preserving hybrid attention-based random forest (ABRF) with stacked bidirectional gated recurrent unit (stacked Bi-GRU) model named hybrid ABRF-Stacked Bi-GRU is proposed in the cloud environment. This proposed model is used to preserve medical data in the cloud from private data leakage. The attention weights along with the trainable parameters are assigned by the ABRF model to the decision trees. The privacy leakage is reduced by combining Jarratt butterfly optimization algorithm with the Ebola optimization search algorithm. However, in various existing methods, the differential privacy model is applied but it does not minimize the efficiency. In such cases it leads to generate data loss and privacy is not enhanced. Data securing in a cloud environment is a complex task. So in order to improve the efficiency the novel hybrid ABRF-Stacked Bi-GRU is proposed to secure the medical data. This proposed method diminished the leakage of data and also effectively improve security and privacy. The ensemble of the ABRF and stacked Bi-GRU model enhances the security of data in a cloud environment. The proposed model is evaluated using Diabetes UCI Dataset and Obesity UCI ML Dataset. The obtained outcomes compared with existing methods showed that the proposed method achieves a high anonymization degree with less Information Loss and takes less time for computation and minimizes the complexity during computation. [ABSTRACT FROM AUTHOR]

Since there has been concern about food security, accurate prediction of wheat yield prior to harvest is a key component. Random Forest (RF) has been used in many classification and regression applications, such as yield estimation, and the performance of RF has improved by tuning its hyperparameters. In this paper, different changes are made to traditional RF for yield estimation, and the performance of RF is evaluated. Accordingly, RFs constructed using various weak learners, as well as a combined RF consisting of different weak learners are assessed by growing weak Gaussian Process Regression (GPR), Decision Tree (DT), Neural Network (NN), and Stepwise Regression (SW) models in the forest. The input data to DTs are also partitioned into leaves in N (e.g., two) dimensional feature space by using clustering in each parent node. In addition, a subset of the training set is randomly sampled with replacement for training a learner in the forest, instead of randomly sampling the whole training set in traditional RF. Using clustering in DTs added flexibility while utilizing NN as a weak learner yielded the most favorable outcomes in our research. The number of input training samples (Itree) to each tree was also identified as a new hyperparameter to the forest, and the prediction results were more influenced by the Itree compared to the known hyperparameters, such as the number of trees in the forest (ntree) and the number of features for each tree (mtry). [ABSTRACT FROM AUTHOR]

This paper analyzes the efficiency of various machine learning models (artificial neural networks, random forest, decision tree, AdaBoost and XGBoost) against the evolution of VBA-based (Visual Basic for Applications) malware over a large period of time (1995–2021). The file set used in our research is comprehensive—approximately 1.9 million files (out of which 944,595 are malicious and the rest are benign)—which allowed to gain insights on the resilience of various machine learning models against the diversity and the evolution of file features that reflect obfuscation techniques in VBA-based malware. In studying detection of VBA-based malware, we focus on characteristics of both the classifiers—proactivity (short-term detection efficiency against future malware), endurance (long-term detection robustness)—and of the detection-wise relevant file features—feature perishability (dynamics of feature relevance). We also describe in some detail—as a prerequisite of the study—various obfuscation techniques used by the malware under investigation during the last decade. [ABSTRACT FROM AUTHOR]

Landslide is one of the most highly frequent natural hazards that can bring serious casualties. One of the most susceptible landslide regions in Indonesia is Bogor area (the Regency and City of Bogor), which records the highest landslide events in the Province of West Java, Indonesia. An assessment of landslide susceptibility is one of the mitigation measures that can spatially model the zone of landslide hazard. Recently, the Landslide Susceptibility Mapping (LSM) model has been developed using Machine Learning (ML) algorithms. However, there is still no agreement yet on which ML technique is the most appropriate for LSM. Accordingly, this paper aims to explore and compare the 7 ML algorithms for generating the most promising LSM. The LSM uses the available 13 landslide causal factors and a dataset consisting of 822 authorized landslide records and 822 prepared non-landslide points. The resulting LSMs are classified into 5 susceptibility levels, and evaluated through the Area Under Curve (AUC) of the Receiver-Operating Curve (ROC) and statistical indices (sensitivity, specificity, precision, F1-score, and accuracy). The resulting LSMs present that: (1) the very high (VH) class has the largest area percentage in all LSM models, (2) generally, the 7 MLs perform excellent for achieving > 90% AUC value, except for the Decision Tree (DT) (87.68%) in model classification, and (3) moreover, the overall accuracy (ACC) reflects that Random Forest (RF) outperforms the other MLs in model prediction. With this promising result, ML-based LSM models can be promoted as one of the mitigation measures for landslide disaster management. [ABSTRACT FROM AUTHOR]

SpMV is a vital computing operation of many scientific, engineering, economic and social applications, increasingly being used to develop timely intelligence for the design and management of smart societies. Several factors affect the performance of SpMV computations, such as matrix characteristics, storage formats, software and hardware platforms. The complexity of the computer systems is on the rise with the increasing number of cores per processor, different levels of caches, processors per node and high speed interconnect. There is an ever-growing need for new optimization techniques and efficient ways of exploiting parallelism. In this paper, we propose ZAKI, a data-driven, machine-learning approach and tool, to predict the optimal number of processes for SpMV computations of an arbitrary sparse matrix on a distributed memory machine. The aim herein is to allow application scientists to automatically obtain the best configuration, and hence the best performance, for the execution of SpMV computations. We train and test the tool using nearly 2000 real world matrices obtained from 45 application domains including computational fluid dynamics (CFD), computer vision, and robotics. The tool uses three machine learning methods, decision trees, random forest, gradient boosting, and is evaluated in depth. A discussion on the applicability of our proposed tool to energy efficiency optimization of SpMV computations is given. This is the first work where the sparsity structure of matrices have been exploited to predict the optimal number of processes for a given matrix in distributed memory environments by using different base and ensemble machine learning methods. [ABSTRACT FROM AUTHOR]

Predicting traffic accident duration is necessary for ensuring traffic safety. Several attempts have been made to achieve high prediction accuracy, but researchers have not considered traffic accident text data in much detail. The limited text data of the first report on an incident describes the characteristics of an accident that are initially available. This paper uses text data fusing and ensemble learning algorithms to build a model to predict an accident's duration, and a preprocessing scheme of accident duration text data is established. Next, the random forest (RF) algorithm is applied to select feature variables of text data related to the traffic incident duration. Last, a text feature vector is introduced to models such as decision tree, k nearest neighbor, support vector regression, random forest, Gradient Boosting Decision Tree, and Xtreme Gradient Boosting. Our results show that the improved RF model has good prediction accuracy with RMSE, MAPE and R2. From this, the textual factors important to determining the duration of the accident are identified. Further, we investigated that the cumulative importance of 60% is sufficient for traffic accident prediction using text data. These results provide insights into minimizing traffic congestion related to accidents and contribute to the input optimization in text prediction. [ABSTRACT FROM AUTHOR]

Drilling costs and time are directly affected by the rate of penetration (ROP). Gas/oil well drilling in challenging environments is a costly operation due to the high cost of wellbore construction, and drilling optimization requires real-time decision-making. Therefore, it is necessary to estimate ROP accurately and thus assist in optimizing the drilling process. In this study, we used total depth, weight on hook, bit, revolutions per minute, torque, standpipe pressure, flow rates, mud weight, mud viscosity, hole size, and formation hardness as input parameters to forecast ROP. Nine machine learning algorithms, i.e., multiple linear regression (MLR), lasso regression, ridge regression, decision tree (DT), bootstrap aggregating (Bagging), random forest (RF), gradient boosting machines (GBM), symbolic regression (SR), and artificial neural networks (ANN) were performed using a dataset consists of 3484 observations. Moreover, several network hyperparameters were adjusted, indicating that using a deep artificial feed-forward network, i.e., [9–11-1] neurons with logarithmic sigmoid transfer function as a predictor model, provided the most accurate model. Results showed that the random forest and artificial neural network models have the highest accuracy, and it estimates ROP with correlation coefficients of 0.93. Overall, the methodology employed in this paper performed well in predicting ROP in the absence of some unavailable parameters such as UCS. Highlights: • Effect of weight on hook (WOH) on rate of penetration (ROP) investigated for the first time by means of multiple (nine) robust machine-learning algorithms. • Multiple linear regression (MLR), lasso regression, ridge regression, decision tree (DT), bootstrap aggregating (Bagging), random forest (RF), gradient boosting machines (GBM), symbolic regression (SR), and artificial neural networks (ANN) intelligent models were proposed for estimating drilling rate of penetration. • The accuracy of the models was ranked as: ANN > RF > GBM > BA > DT > SR > Lasso > MLR > Ridge. • A comprehensive databank of 3484 data points from a gas field in the Zagros basin was used. The predicted rate of penetration is highly influenced by the values of depth, WOH and RPM. [ABSTRACT FROM AUTHOR]

Network Intrusion Detection (NID) is an important domain of research and as the network traffic is growing enormously, new challenges are emerging forth in terms of quality of service and security thus a reliable and robust network intrusion detection system (NIDS) has become an inevitable requirement. Researchers have developed and implemented many rule-based and statistical-based techniques, as well as machine learning (ML) approaches such as K-Nearest Neighbour (K-NN), Decision Tree (DT), Random Forest (RF), Multi-Layer Perceptron (MLP) and Naive-Bayes for this pursuit. But, a method is still in need that entails certain required characteristics like low computational cost, high classification accuracy, able to deal efficiently with large datasets having redundant data and high dimensionality. In this paper, a novel method is proposed that includes all these characteristics. Further, the proposed method is based on the rule induction technique using separate and-conquer algorithm known as CN2 for a network intrusion detection system. The performance of the proposed CN2 rule induction algorithm is also compared experimentally with K-NN, DT, RF, MLP and Naive-Bayes classifiers for the Kddcup99 dataset and it has been observed that the performance of the CN2 rule induction algorithm is better in comparison to other classifiers considered in this experimentation. [ABSTRACT FROM AUTHOR]

Economic progress has been founded on environmental pressure by generating all types of environmental harm, such as an increase in greenhouse gases in the atmosphere and significant climate change, which has forced people to reflect. Building a people-oriented, comprehensive, coordinated system to compile experience and lessons assists the government in formulating appropriate policies according to the meaning of the scientific development viewpoint. Based on the above, this research designs Cost–Benefit Decision Tree (CBDT) Algorithm for the economic protection development level using a Decision Tree (DT) and Ensemble Algorithm. In the design of ML Algorithm, this research work initially introduces the Decision Tree model for economic protection evaluation by emphasizing its value as a basic machine-learning framework for fully examining many aspects of economic protection development. Second, it investigates the critical role of Ensemble Algorithms, such as Random Forest in augmenting the capabilities of DT. Third, the paper focuses on the development of a complete Economic Protection Evaluation Index System using the DT Algorithm. To provide a robust, scientifically sound, and practicable framework for evaluating economic protection, this approach integrates scientificity, comparability, systematization, hierarchy, relative decoupling, and data availability principles. Finally, the study constructs the suggested ML Algorithm using the insights from the preceding parts. This algorithm combines the Decision Tree and Random Forest models to generate a single, effective tool for evaluating economic development levels. The experimental findings suggest that the proposed hybrid assessment technique is effective across several dimensions. The proposed CBDT demonstrated its capacity to adapt to changing data circumstances in evaluating low-carbon efficiency using real-world data from 30 major Chinese cities. Furthermore, the algorithm regularly beats rival techniques regarding efficiency improvements, with an average improvement of 12.5% across all cities. Notably, it outperforms other algorithms regarding computational efficiency, with a quicker execution time and an exceptional accuracy rate of 93.2%. The proposed ML Algorithm improves decision-making in the economic protection development level field by effectively balancing cost and benefit considerations. It will improve accuracy and efficiency to outperform existing approaches, thereby advancing economic protection evaluation and decision-making processes in this domain. [ABSTRACT FROM AUTHOR]

In the last few decades, technological improvements have had a considerable impact on education, much like many other fields of society and human endeavor. The applications of today's cutting-edge technology are numerous in education, but recently researchers focused on using these technologies, especially artificial intelligence (AI) and machine learning (ML), for students' performance prediction prior to the exam. It makes perfect sense to forecast student success to help all participants in the educational process. Student performance prediction may assist them in the selection of suitable courses and in creating their academic schedules. Keeping in mind the importance of student performance evaluation, this paper analyzes the prediction rate of college English level 4 by using one of the powerful ML algorithms called random forest (RF). RF uses numerous classifiers, or "ensembles", rather than just one classifier, and is based on the decision tree technique. To do this, we constructed input and output variables and collected data for these variables, including basic student information (gender, ethnicity, major), English scores on college admission exams, college English scores (total of four semesters), and extracurricular activities of college students. Preprocessing was performed on the collected data, which included the removal of unnecessary attributes, handling outliers, normalization, and data cleaning. After the preprocessing, the features were extracted and transformed into reduced dimensionality by the local-preserving projection (LPP) algorithm. From the extracted features, we selected only the most relevant in order to feed them as input to the RF model. The model is implemented in MATLAB in order to evaluate its performance. The efficiency of the proposed model is evaluated with the help of experiments in order to verify the effectiveness of the model. The performance of the RF algorithm-based college English IV pass rate prediction model is evaluated by computing the prediction accuracy, recall rate, and hit rate of the classification results. We achieved a prediction accuracy of 96.5%, a recall rate of 89.5%, and a hit rate of 93.3%. The results show that the random RF-based prediction model for college English level 4 has a good classification effect and that the prediction results are more accurate. [ABSTRACT FROM AUTHOR]

Cloud failure is one of the critical issues since it can cost millions of dollars to cloud service providers, in addition to the loss of productivity suffered by industrial users. Fault tolerance management is the key approach to address this issue, and failure prediction is one of the techniques to prevent the occurrence of a failure. One of the main challenges in performing failure prediction is to produce a highly accurate predictive model. Although some work on failure prediction models has been proposed, there is still a lack of a comprehensive evaluation of models based on different types of machine learning algorithms. Therefore, in this paper, we propose a comprehensive comparison and model evaluation for predictive models for job and task failure. These models are built and trained using five traditional machine learning algorithms and three variants of deep learning algorithms. We use a benchmark dataset, called Google Cloud Traces, for training and testing the models. We evaluated the performance of models using multiple metrics and determined their important features, as well as measured their scalability. Our analysis resulted in the following findings. Firstly, in the case of job failure prediction, we found that Extreme Gradient Boosting produces the best model where the disk space request and CPU request are the most important features that influence the prediction. Second, for task failure prediction, we found that Decision Tree and Random Forest produce the best models where the priority of the task is the most important feature for both models. Our scalability analysis has determined that the Logistic Regression model is the most scalable as compared to others. [ABSTRACT FROM AUTHOR]

Formation resistivity (Rt) is a vital property for formation evaluation and calculation of water saturation and hydrocarbon in places. Rt can be estimated using core analysis and well logging. However, these processes are expensive and time-consuming. In addition, due to tool failure, and poor wellbore conditions, part of the well logging records may be missed. Hence, the objective of this paper is to predict the true formation resistivity in complex carbonate sections using decision tree (DT) and Random Forests (RF) machine learning (ML) techniques as a function of available well logging data. A data set of 5500 data points were collected from two vertical wells in carbonate formation. The data includes gamma-ray, bulk density, neutron density, compressional wave transit time, shear transient time, and the corresponding Rt. Data from Well-1 were used to develop the DT and RF models with training to the testing splitting ratio of 70:30. Dataset from Well-2 was used to validate the optimized models. The results showed the capabilities of the ML models to predict the formation resistivity from well-logging data. The correlation coefficient (R) between the actual and the predicted output values and the root mean square error (RMSE) was used to evaluate the models performance. R value for the RF model was found to be 0.99, and 0.98 for the training and the testing stages with a validation R value of 0.94. The RMSE for the developed models was less than 0.38 for training, testing, and validation stages. Using ML to predict the formation resistivity can fill the missing gaps in log tracks and save money by removing resistivity logs running in all offset wells in the same field. [ABSTRACT FROM AUTHOR]

Author profiling is the process of analysing textual data to extract information about various personality traits of the author. It has both commercial and social implications. Many approaches have been proposed in past to increase the accuracy of the extracted information. In this paper, we applied natural language processing and machine-learning approach for author profiling. NLP techniques like Tokenization, lemmatization, word and char n-grams are used in integration with machine learning classifiers like logistic regression (LR), random forest (RF), decision tree (DT) and support vector machine (SVM). The proposed method obtained an accuracy of 81.2%, 79.8%, 63.2% and 88.0% for the four classifiers respectively for gender prediction and 72.5%, 68.1%, 53.7% and 81.0% respectively for age prediction i.e. SVM outperformed the other classifiers with an accuracy of 88.0% for gender prediction and 81.0% for age prediction. [ABSTRACT FROM AUTHOR]

Predictive maintenance employing machine learning techniques and big data analytics is a benefit to the industrial business in the Industry 4.0 era. Companies, on the other hand, have difficulties as they move from reactive to predictive manufacturing processes. The purpose of this paper is to demonstrate how data analytics and machine learning approaches may be utilized to predict production delays in a quarry firm as a case study. The dataset contains production records for six months, with a total of 20 columns for each production record for two machines. Cross Industry Standard Process for Data Mining approach is followed to build the machine learning models. Five predictive models were created using machine learning algorithms such as Decision Tree, Neural Network, Random Forest, Nave Bayes and Logistic Regression. The results show that Multilayer Perceptron Neural Network and Logistic Regression outperform other techniques and accurately predicts production delays with a F-measure score of 0.973. The quarry company's improved decision-making reducing potential production line delays demonstrates the value of this study. [ABSTRACT FROM AUTHOR]

Predictive maintenance refers to predicting malfunctions using data from monitoring equipment and process performance measurements. Machine learning algorithms and techniques are often used to analyze equipment monitoring data. Machine learning is the process in which a computer can work more precisely by collecting and analyzing data. It is often the case that machine learning algorithms use supervised learning, in which labelled data is used to feed the algorithm. However, there are many supervised machine learning algorithms available. Therefore, choosing the best-supervised machine learning algorithm to resolve predictive maintenance issues is not trivial. This paper aims to increase the performance of predictive maintenance and achieve its goals by selecting the most suitable supervised machine learning algorithm. Based on the most commonly used criteria in research articles, we selected three supervised machine learning algorithms from a comparative study: Random forest, Decision tree and KNN. We then tested selected algorithms on data from real-world and simulation scenarios. Finally, we conducted the experiment based on vibration analysis and reliability evaluation. We noticed that Random forests and Decision trees obtained slightly the same performance. KNN is a better classification algorithm for extensive volumes of data; on the contrary, Random forest performs better in the case of small datasets. [ABSTRACT FROM AUTHOR]

An adversary can use SSH communication as a route for information leakage or hacking. Many studies have focused on TCP header analysis to detect encrypted communication. However, SSH detection using TCP header analysis is limited when changing TCP port information or modifying components of the SSH protocol. Various machine-learning (ML) techniques have been introduced to enhance network traffic classification by analyzing TCP headers. Most ML-based traffic classification research has analyzed network packet flows. However, because of the complex structures and the various implementations of the TCP protocol, a lot of time and resources are required for the recombination of network packet flows. This paper presents a novel contribution to overcome the problems of network packet analysis that employs web proxy session logs, which do not require the recombination of packets to prepare a dataset for analysis. Moreover, we propose a hybrid predictive model that is useful for web proxy session log analysis. In the modeling process, we collected the web proxy logs from an actual network of ICT companies (more than 10,000 employees, Seoul, South Korea) and used the random forest and decision tree algorithms for the supervised learning. The detection rate (DR) for the training dataset was 99.9%, which is similar to or higher than that of other studies using ML and deep learning. Using the dataset of DARPA99, we proved that the DR and FPR for our proposed model were better than those achieved by Alshammari et al.'s model. We expect that the proposed predictive model can be used to block illegal attempts at SSH communication over HTTP CONNECT by changing the destination port and to detect novel illegal communication protocols. [ABSTRACT FROM AUTHOR]

As the number of smartphone users is increasing exponentially, there is an increase in the availability of continuous sensor data, which has attracted enormous interest in sensor-based human-activity recognition (HAR). Recognizing human activities is particularly important in detecting abnormal activities and tracking a person's physical activity, especially in healthcare applications, among many others. In this paper, HAR analysis is conducted with three different machine learning algorithms (Support Vector Machines (SVM), Decision tree, and random forest methods) based on smartphone sensors. Machine learning algorithms are capable of identifying and differentiating between different human activities using mobile phone sensor data. The smartphone sensors (gyroscope and accelerometer) data are recorded at the Koneru Lakshmaiah Education Foundation University campus, Guntur, India, with different human activities. In this research work, the data from smartphone mobile sensors were initially analysed with SVM, decision tree, and random forest algorithms. To evaluate the machine learning algorithm's accuracy, F1 score for different smartphone sensors for both individually and combined is estimated. The results indicate that the proposed machine learning methods can derive a relation between type of activity, algorithm, smartphone sensors data, and their corresponding accuracy. The outcome of this work would be beneficial for detecting abnormal features of older people with a smartphone device. [ABSTRACT FROM AUTHOR]

Recently, Bayesian Optimization (BO) provides an efficient technique for selecting the hyperparameters of machine learning models. The BO strategy maintains a surrogate model and an acquisition function to efficiently optimize the computation-intensive functions with a few iterations. In this paper, we demonstrate the utility of the BO to fine-tune the hyperparameters of a Random Forest (RF) model for a problem related to the recognition of splice-junction genetic sequences. Locating these splice-junctions prompts further understanding of the DNA splicing process. Specifically, the BO algorithm optimizes four RF hyperparameters: number of trees, number of splitting features, splitting criterion, and leaf size. The optimized RF model automatically selects the most predictive features of the training data. The dataset is obtained from the UCI machine learning repository where half of the records represent two different types of splice-junctions and the other half does not represent any splice-junction. Experimental results proved the advantage of the BO-RF with 99.96% and 97.34% training and test classification accuracies respectively. The results also demonstrated the ability of the RF model to select the most important features, ensuring the best possible results using Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and decision tree (DT) models. Some practical procedures in model development and evaluation such as out-of-bag error and cross-validation approaches are also referred to. [ABSTRACT FROM AUTHOR]

For a long time, elucidating the tectonic setting of unknown rock samples has been a focal point for geologists. Traditional methodologies for this purpose have been scrutinized increasingly due to their inherent limitations. In response to these challenges, this paper applies modern machine learning techniques to analyze the geochemical data of igneous rocks and improve understanding of tectonic settings. By employing a variety of machine learning models, including Decision Trees, K-Nearest Neighbors, Support Vector Machines, Random Forests, Extreme Gradient Boosting, and Artificial Neural Networks, and training with 23 features comprising nine major elements (SiO2, TiO2, Al2O3, CaO, MgO, MnO, Na2O, K2O, and P2O5) along with 14 trace elements (La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu), the study successfully distinguished between seven different tectonic settings. Among these models, Random Forest, Extreme Gradient Boosting, and Artificial Neural Networks demonstrated superior classification accuracy and recall rates, with accuracies of 0.85, 0.87, and 0.86, respectively. This validates the effectiveness and potential of machine learning technologies in distinguishing the tectonic settings of igneous rocks through their geochemical elements. To enable geologists and researchers to more accurately understand and predict the origins of igneous rocks without the need to master machine learning knowledge, a user-friendly software, GeoTectAI, has been developed. [ABSTRACT FROM AUTHOR]

This study evaluated the performance of multiple models that used machine learning to anticipate wind speed in the city of Dhaka. The NASA Power website provided the data set for this investigation. The models used for prediction included the decision tree regressor, support vector regressor, random forest, linear regression, neural network and polynomial regression. A hold-out check and k-fold cross-validation were used to assess how well these models performed. With the highest R2 scores and lowest RMSEs on both the validation and test sets, the results demonstrated that the polynomial regression model performed the best. With the lowest R2 scores and largest RMSEs on both sets, the decision tree model scored the poorest. High R2 scores and low RMSEs were achieved by the random forest model, which had comparable performance to the polynomial regression model but required a longer computation time. In addition, the neural network model demonstrated commendable predictive accuracy, yielding an R2 score of 0.67 and a low RMSE of 0.57. However, its application is contingent on the availability of substantial computational resources, given its extensive computation time of 457.93 s. The study concludes by highlighting the efficacy of the Polynomial Regression model as the optimal choice for wind speed prediction in Dhaka, offering a balance between superior performance and efficient computation. This insight provides valuable guidance for practitioners and researchers seeking effective models for similar applications. [ABSTRACT FROM AUTHOR]

Machine learning is frequently used in various geotechnical applications nowadays. This study presents a statistics and machine learning model for settlement prediction of helical piles that relates compressive service load and soil parameters as a group with the pile parameters. Machine learning algorithms such as Decision Trees, Random Forests, AdaBoost, and Artificial Neural Networks (ANN) were used to develop the predictive models. The models were validated using cross-validation techniques and tested on an independent dataset to assess their accuracy and generalizability. Numerical investigation is used here to supplement the field data by simulating various soil conditions and pile geometries that have not been tested in the field. This study compiled numerical results of 3600 models. As the models are well-calibrated and validated, the data from these models can be reasonably assumed to simulate the ground situation. At the end of this study, a comparative analysis of statistic learning and machine learning (ML) was done using the field axial load tests database and numerical investigation on helical piles. It is observed that ML models like Decision Trees and Random Forests provided the better model with R-squared values of 0.92 and 0.96, respectively, for large diameters. The authors believe this study will permit engineers and state agencies to understand this prediction model's efficacy better, resulting in a more resilient approach to designing large-diameter helical piles for the compressive load. [ABSTRACT FROM AUTHOR]

The escalating frequency and sophistication of cyber threats pose significant challenges to traditional intrusion detection methods. Signature-based misuse detection, hybrid detection, and anomaly detection, while effective in isolation, often struggle to keep pace with the ever-evolving tactics employed by attackers. This research stems from the pressing need to develop intrusion detection models that offer improved accuracy, adaptability, and robustness in the face of rapidly changing attack vectors. Our study leverages a comprehensive set of experiments conducted on diverse datasets, including N-BaIoT, NSL-KDD, and CICIDS2017. The primary focus is on the evaluation and comparison of machine learning algorithms such as Random Forest, XGBoost, and decision trees. Notably, our research culminates in the development of a hybrid intrusion detection model that combines the strengths of these algorithms. The results of our experiments indicate that the hybrid model, particularly when combining Random Forest and XGBoost, exhibits exceptional accuracy. This approach outperforms individual algorithms, achieving an accuracy rate of 97% in certain cases. We attribute this success to the ensemble learning approach, which capitalizes on the consensus of diverse classifiers. In conclusion, our research not only demonstrates the effectiveness of ensemble learning in enhancing intrusion detection but also underscores the importance of continuous adaptation in the face of evolving threats. By leveraging network traffic data, our hybrid model offers a promising avenue for bolstering intrusion classification in Big Data environments. [ABSTRACT FROM AUTHOR]

In this work, our proof-of-concept study can be used to predict the number of cells within printed droplets based on droplet velocity at two different points along the nozzle-substrate distance using machine learning approaches. A novel high-throughput contactless method that combines the use of an optical system and machine learning algorithms was utilized for various applications such as cell detection within single droplets (presence/absence of cells) and prediction of the total number of printed cells within multiple droplets by measuring the droplet deceleration between two positions along the nozzle-substrate distance. The proposed method in this work has demonstrated good accuracy in cell prediction within single droplet (presence/absence of cells) and low prediction error in determining number of cells within multiple droplets by reducing the error by a factor of for N droplets measured in a batch. The performance of five different machine learning algorithms such as linear regression, support vector regression, decision tree regressor, random forest regression, and extra tree regression were compared to determine the best algorithm for each type of application. The random forest regressor algorithm demonstrated the highest accuracy at 80% in cell prediction (presence/absence of cells) within single droplets, while the extra tree regressor demonstrated the lowest mean error of 12% in predicting the number of printed cells within multiple droplets (e.g., 20 droplets on same spot). By incorporating these models in a droplet monitoring system, live assessment of the number of printed cells during an inkjet-based bioprinting process can be achieved. [ABSTRACT FROM AUTHOR]

This article aims to predict the mid-span displacements of reinforced concrete (RC) columns exposed to blast loading with machine learning models. Machine learning models including the gradient boosting decision tree (GBDT) model and the random forest (RF) model are developed. The model hyperparameters in the machine learning models are globally optimized using Bayesian optimization method. The dataset used to train and test the models is constructed by collecting data from shock-tube-simulated experiments. Results show that the performance of the optimized models is significantly improved. By comparing the performance metrics of two models, it is evident that the GBDT model is superior to the RF model in predicting mid-span displacements. The optimized GBDT model is then used to predict the displacements of columns during field tests, and the experimental results and predictive results are relatively consistent, which verifies the effectiveness of GBDT model with Bayesian optimization in damage assessment for RC columns. [ABSTRACT FROM AUTHOR]

Predicting oil well behavior regarding the integrity of its equipment during production and anticipating behavioral changes and anomalies are among the main challenges in oil production. In this context, this study focuses on the development of predictive models for real-time monitoring of well behavior using sensor data from production wells. An unsupervised Novelty and Outlier Detection model has been introduced with a specific focus on predicting instances of unexpected subsurface safety valve closures in subsea wells. This model effectively classifies anomalies observed in these systems by leveraging real-world pressure and temperature data sourced from published literature. The methodology involves the implementation of a floating window for assembling training and test sets. Additionally, a comprehensive investigation is conducted into the impact of hyperparameters and the model's threshold value (cp threshold). The results highlight the effectiveness of the developed model, observed through the accuracy achieved around 99.9% in predicting spurious closure events of the Downhole Safety Valve. On the same dataset, previous works reported 99.9% accuracy by using long short-term memory (LSTM) autoencoder, 87.1% by using random forest, and 60% with the Decision Tree method. Looking at F1-SCORE values, the developed model performs the best, followed by the LSTM model, both of which are significantly superior to the Decision Tree and random forest models. Furthermore, the model's applicability is validated through testing in ultradeep water subsea wells within the pre-salt area of the Santos Basin. The significance lies in the potential for this research to enhance anomaly prediction in offshore wells, consequently reducing the costly interventions due to equipment malfunctions. Timely detection and corrective actions, facilitated by the model, can mitigate production loss and safeguard well integrity, addressing critical concerns in the oil and gas industry. [ABSTRACT FROM AUTHOR]

Ensemble methods are supervised learning algorithms that provide highly accurate solutions by training many models. Random forest is probably the most widely used in regression and classification problems. It builds decision trees on different samples and takes their majority vote for classification and average in case of regression. However, such an algorithm suffers from a lack of explainability and thus does not allow users to understand how particular decisions are made. To improve on that, we propose a new way of interpreting an ensemble tree structure. Starting from a random forest model, our approach is able to explain graphically the relationship structure between the response variable and predictors. The proposed method appears to be useful in all real-world cases where model interpretation for predictive purposes is crucial. The proposal is evaluated by means of real data sets. [ABSTRACT FROM AUTHOR]

The complex trajectories of higher education students are deviations from the regular path due to delays in completing a degree, dropping out, taking breaks, or changing programmes. In this study, we investigated degree changing as a cause of complex student trajectories. We characterised cohorts of students who graduated with a complex trajectory and identified the characteristics that influenced the time to graduation. To support this predictive task, we employed machine learning techniques such as neural networks, support vector machines, and random forests. In addition, we used interpretable techniques such as decision trees to derive managerial insights that could prove useful to decision-makers. We validated the proposed methodology taking the University of Porto (Portugal) as case study. The results show that the time to degree (TTD) of students with and without complex trajectories was different. Moreover, the proposed models effectively predicted TTD, outperforming two benchmark models. The random forest model proved to be the best predictor. Finally, this study shows that the factors that best predict TTD are the median TTD and the admission regime of the programme of destination of transfer students, followed by the admission average of the previous programme. By identifying students who take longer to complete their studies, targeted interventions such as counselling and tutoring can be promoted, potentially improving completion rates and educational outcomes without having to use as many resources. [ABSTRACT FROM AUTHOR]

What dataset features affect machine learning (ML) performance has primarily been unknown in the current literature. This study examines the impact of tabular datasets' different meta-level and statistical features on the performance of various ML algorithms. The three meta-level features this study considered are the dataset size, the number of attributes and the ratio between the positive (class 1) and negative (class 0) class instances. It considered four statistical features for each dataset: mean, standard deviation, skewness and kurtosis. After applying the required scaling, this study averaged (uniform and weighted) each dataset's different attributes to quantify its four statistical features. We analysed 200 open-access tabular datasets from the Kaggle (147) and UCI Machine Learning Repository (53) and developed ML classification models (through classification implementation and hyperparameter tuning) for each dataset. Then, this study developed multiple regression models to explore the impact of dataset features on ML performance. We found that kurtosis has a statistically significant negative effect on the accuracy of the three non-tree-based ML algorithms of the Support vector machine (SVM), Logistic regression (LR) and K-nearest neighbour (KNN) for their classical implementation with both uniform and weighted aggregations. This study observed similar findings in most cases for ML implementations through hyperparameter tuning, except for SVM with weighted aggregation. Meta-level and statistical features barely show any statistically significant impact on the accuracy of the two tree-based ML algorithms (Decision tree and Random forest), except for implementation through hyperparameter tuning for the weighted aggregation. When we excluded some datasets based on the imbalanced statistics and a significantly higher contribution of one attribute compared to others to the classification performance, we found a significant effect of the meta-level ratio feature and statistical mean and standard deviation features on SVM, LR and KNN accuracy in many cases. Our findings open a new research direction in understanding how dataset characteristics affect ML performance and will help researchers select appropriate ML algorithms for a possible optimal accuracy outcome. [ABSTRACT FROM AUTHOR]

The escalating threat of natural disasters to public safety worldwide underlines the crucial role of effective environmental risk management tools, such as insurance. This is particularly evident in the case of earthquakes that occurred in Oklahoma between 2011 and 2020, which were linked to wastewater injection, underscoring the need for earthquake insurance. In this regard, from a survey of 812 respondents in Oklahoma, USA, we used supervised machine learning techniques (i.e., logit, ridge, least absolute shrinkage and selection operator (LASSO), decision tree, and random forest classifiers) to identify the factors that influence earthquake insurance uptake and to predict individuals who would acquire earthquake insurance. Our findings reveal that influential factors that affect earthquake insurance uptake include demographic factors such as older age, male gender, race, and ethnicity. These were found to significantly influence the decision to purchase earthquake insurance. Additionally, individuals residing in rental properties were less likely to purchase earthquake insurance, while longer residency in Oklahoma had a positive influence. Past experience of earthquakes was also found to positively influence the decision to purchase earthquake insurance. Both decision trees and random forests demonstrated good predictive capabilities for identifying earthquake insurance uptake. Notably, random forests exhibited higher precision and robustness, emerging as an encouraging choice for earthquake insurance modeling and other classification problems. Empirically, we highlight the importance of insurance as an environmental risk management tool and emphasize the need for awareness and education on earthquake insurance as well as the use of supervised machine learning algorithms for classification problems. [ABSTRACT FROM AUTHOR]

What causes viewers to change their winner perception during a televised debate? The article addresses this question, drawing on a large-N field study of the 2021 chancellor debate in Germany, which contains survey and real-time response data for 4613 participants. Using machine learning techniques, we identify determinants of why participants change their opinion about who is winning the discussion during the debate. Our analysis based on random forest and decision tree models shows in detail, first, what factors drive debate winner perceptions in the course of televised debate reception. Second, we reveal what combinations of political predispositions and candidate statements are necessary to change the viewers' debate winner perception. In doing so, third, we expand the toolbox of empirical debate research with our analysis based on machine learning algorithms. Our findings indicate that pre-debate chancellor preference and candidate images play a crucial role in determining post-debate perception change, while party identification is less important in predicting changes. We can also directly identify several speech moments in the debate that shifted viewer perception, a novel approach to evaluating political debate performance. [ABSTRACT FROM AUTHOR]

The dimensions of microarray datasets are very large, containing noise and redundancy. The problem with microarray datasets is the presence of more features compared to the number of samples, which adversely affects algorithm performance. In other words, the number of columns exceeds the number of rows. Therefore, to extract precise information from microarray datasets, a robust technique is required. Microarray datasets play a critical role in detecting various diseases, including cancer and tumors. This is where feature selection techniques come into play. In recent times, feature selection (FS) has gained significant importance as a data preparation method, particularly for high-dimensional data. It is preferable to address classification problems with fewer features while maintaining high accuracy, as not all features are necessary to achieve this goal. The primary objective of feature selection is to identify the optimal subset of features. In this context, we will employ the Differential Evolution (DE) algorithm. DE is a population-based stochastic search approach that has found widespread use in various scientific and technical domains to solve optimization problems in continuous spaces. In our approach, we will combine DE with three different classification algorithms: Random Forest (RF), Decision Tree (DT), and Logistic Regression (LR). Our analysis will include a comparison of the accuracy achieved by each algorithmic model on each dataset, as well as the fitness error for each model. The results indicate that when feature selection was used the results were better compared to the results where the feature selection was not used. [ABSTRACT FROM AUTHOR]

Lithology identification is critical in the interpretation of well-logging data for petroleum exploration and development. However, the limited availability of labeled well-logging data for machine learning model training can lead to compromised accuracy in lithology classification models. Here, we propose a semi-supervised lithology identification model to overcome this challenge. Our framework consists of Bayesian optimization for tuning ensemble algorithms, including random forest, gradient boosting decision tree, extremely randomized trees, and adaptive boosting, to establish a high-quality baseline model for semi-supervised learning. We also employ a self-training strategy to increase the number of labeled samples in the training set and use the predicted label with the highest confidence as a pseudo-label to reduce the accumulation of deviation caused by incorrect pseudo-labels. Our semi-supervised coarse-to-fine framework improves rock classification accuracy, particularly for sandstone. Testing our model on well-logging data from two real regions, we found that the ExtraRF-based semi-supervised model in the HGF area performs the best, with a maximum classification accuracy of 91.6 % , which is 5 % higher than the original coarse-to-fine model without using Bayesian optimization and pseudo-labeling techniques. [ABSTRACT FROM AUTHOR]

Email is a useful communication medium for better reach. There are two types of emails, those are ham or legitimate email and spam email. Spam is a kind of bulk or unsolicited email that contains an advertisement, phishing website link, malware, Trojan, etc. This research aims to classify spam emails using machine learning classifiers and evaluate the performance of classifiers. In the pre-processing step, the dataset has been analyzed in terms of attributes and instances. In the next step, thirteen machine learning classifiers are implemented for performing classification. Those classifiers are Adaptive Booster, Artificial Neural Network, Bootstrap Aggregating, Decision Table, Decision Tree, J48, K-Nearest Neighbor, Linear Regression, Logistic Regression, Naïve Bayes, Random Forest, Sequential Minimal Optimization and, Support Vector Machine. In terms of accuracy, the Random Forest classifier performs best and the performance of the Naïve Bayes classifier is substandard compared to the rest of the classifiers. Random Forest classifier had the accuracy of 99.91% and 99.93% for the Spam Corpus and Spambase datasets respectively. The naïve Bayes classifier had the accuracy of 87.63% and 79.53% for the Spam Corpus and Spambase datasets respectively. [ABSTRACT FROM AUTHOR]

A physiological-based emotion recognition system (ERS) with a unimodal approach such as an electrocardiogram (ECG) is not as popular compared to a multimodal approach. However, a single modality has the advantage of lower development and computational cost. Therefore, this study focuses on a unimodal ECG-based ERS. The ECG-based ERS has the potential to become the next mass-adopted consumer application due to the wide availability of wearable and mobile ECG devices in the market. Currently, ECG-inclusive affective datasets are limited, and many of the existing datasets have small sample sizes. Hence, ECG-based ERS studies are stunted by the lack of quality data. A novel multi-filtering augmentation technique is proposed here to increase the sample size of the ECG data. This technique augments the ECG signals by cleaning the data in different ways. Three small ECG datasets labelled according to emotion state are used in this study. The benefit of the proposed augmentation techniques is measured using the classification accuracy of five machine learning algorithms; k-nearest neighbours (KNN), support vector machine, decision tree, random forest and multilayer perceptron. The results show that with the proposed technique, there is a significant improvement in performance for all the datasets and classifiers. KNN classifier improved the most with the augmented data and the reported classification accuracies of over 90%. [ABSTRACT FROM AUTHOR]