Tree-based Classification Research Articles

Transient classifiers for Fin. Benchmarks for LSST

The upcoming Legacy Survey of Space and Time (LSST) at the Vera C. Rubin Observatory is expected to detect a few million transients per night, which will generate a live alert stream during the entire ten years of the survey. This stream will be distributed via community brokers whose task is to select subsets of the stream and direct them to scientific communities. Given the volume and complexity of the anticipated data, machine learning (ML) algorithms will be paramount for this task. We present the infrastructure tests and classification methods developed within the Fink broker in preparation for LSST. This work aims to provide detailed information regarding the underlying assumptions and methods behind each classifier and enable users to make informed follow-up decisions from Fink photometric classifications. Using simulated data from the Extended LSST Astronomical Time-series Classification Challenge (ELAsTiCC), we showcase the performance of binary and multi-class ML classifiers available in These include tree-based classifiers coupled with tailored feature extraction strategies as well as deep learning algorithms. Moreover, we introduce the CBPF (Centro Brasileiro de Pesquisas F\'isicas) Alert Transient Search (CATS), a deep learning architecture specifically designed for this task. Our results show that Fink classifiers are able to handle the extra complexity that is expected from LSST data. CATS achieved $ 93<!PCT!>$ precision for all classes except `long' (for which it achieved $ 83<!PCT!>$) while our best performing binary classifier achieves $ 98<!PCT!>$ precision and $ 99<!PCT!>$ completeness when classifying the periodic class. ELAsTiCC was an important milestone in preparing the Fink infrastructure to deal with LSST-like data. Our results demonstrate that Fink classifiers are well prepared for the arrival of the new stream, but this work also highlights that transitioning from the current infrastructures to Rubin will require significant adaptation of the currently available tools. This work was the first step in the right direction.

Comprehensive prediction of outcomes in patients with ST elevation myocardial infarction (STEMI) using tree-based machine learning algorithms

ST elevation myocardial infarction (STEMI), a subtype of acute coronary syndrome, is one of the leading causes of morbidity and mortality. Revascularization using primary percutaneous coronary intervention (PPCI) is the gold standard treatment. Despite the restoration of myocardial blood flow, some patients experience adverse outcomes. Early detection of high-risk patients would facilitate timely management, potentially improving their morbidity, mortality, and quality of life. In-depth characterization of the aortic pressure (AP) waveform may identify a high-risk patient cohort. We present tree-based classifiers and features extracted from the AP signals to identify patients at risk of adverse outcomes.This is a single-center, retrospective cohort study that included 605 eligible STEMI patients [64.2 ± 13.2 years, 71.4 % (432) males] treated with PPCI. Outcomes, including mortality (within 30-day and 1-year), and in-hospital events such as prolonged in-hospital stay (>4 days) for medical reasons, a new diagnosis of heart failure (HF), diuretic use for more than 24 h, intubation-ventilation or BiPAP use, and inotropic and/or vasopressor use, were recorded. We extracted features mainly from denoised AP signals recorded during PPCI, followed by different feature selection algorithms and classification methods to predict outcomes. Various classifiers such as tree-based classifiers, including random forest (RF), adaptive boosting (AdaBoost), extreme gradient boosting (XGBoost), and CatBoost, were used.Using recursive feature elimination (RFE) as the feature selection method and the CatBoost classifier, we achieved a receiver operating characteristic curve's area under the curve (AUC) of 80 % for all outcomes except for the new diagnosis of HF and diuretic use (>24 h). For the new diagnosis of HF and diuretic use (>24 h), the AUC values were 73 % and 79 %, respectively.In conclusion, tree-based classifiers using features extracted from AP traces can effectively identify patients at risk of adverse outcomes in patients with STEMI.

Health monitoring of automotive clutch system through feature fusion and application of tree-based classifiers

Clutch systems are crucial components of automotive systems, essential for transferring engine power to the wheels through gear shifts. A failed clutch can halt gear shifts and power transmission, rendering a vehicle immobile. Effective fault diagnosis is vital for ensuring reliability, preventing breakdowns, and addressing root causes promptly. Consequently, condition monitoring of the clutch is essential for maintaining system reliability and minimizing unwanted breakdowns. This paper presents an innovative condition-monitoring technique derived from vibration analysis to detect faults in the clutch system. The study involves extracting statistical, histogram, and autoregressive moving average features from acquired vibration signals. The J48 decision tree algorithm is utilized to select the most significant features, which are then classified using tree-based classifiers across three load conditions (no load, 5 kg, and 10 kg) and six clutch conditions. The experimental results demonstrate that the feature fusion strategy significantly enhances classification accuracy. The obtained results state that the classification accuracies for no load, 5 kg load, and 10 kg load were 98.33%, 100.00%, and 99.16%, respectively, while applying feature fusion strategies. This study highlights the effectiveness of feature fusion in improving fault diagnosis accuracy for clutch systems, presenting a robust method for real-time condition monitoring in automotive applications.

Leveraging machine learning for the detection of structured interference in Global Navigation Satellite Systems

Radio frequency interference disrupts services offered by Global Navigation Satellite Systems (GNSS). Spoofing is the transmission of structured interference signals intended to deceive GNSS location and timing services. The identification of spoofing is vital, especially for safety-of-life aviation services, since the receiver is unaware of counterfeit signals. Although numerous spoofing detection and mitigation techniques have been developed, spoofing attacks are becoming more sophisticated, limiting most of these methods. This study explores the application of machine learning techniques for discerning authentic signals from counterfeit ones. The investigation particularly focuses on the secure code estimation and replay (SCER) spoofing attack, one of the most challenging type of spoofing attacks, ds8 scenario of the Texas Spoofing Test Battery (TEXBAT) dataset. The proposed framework uses tracking data from delay lock loop correlators as intrinsic features to train four distinct machine learning (ML) models: logistic regression, support vector machines (SVM) classifier, K-nearest neighbors (KNN), and decision tree. The models are trained employing a random six-fold cross-validation methodology. It can be observed that both logistic regression and SVM can detect spoofing with a mean F1-score of 94%. However, logistic regression provides 165dB gain in terms of time efficiency as compared to SVM and 3 better than decision tree-based classifier. These performance metrics as well as receiver operating characteristic curve analysis make logistic regression the desirable approach for identifying SCER structured interference.

Processing and Comparison of GBoost, XGBoost, and Random Forest in Titanic Survival Prediction

Abstract. Contemporarily, the machine learning has evolved from its early concepts to a sophisticated field consisting of advanced algorithms and diverse applications. Tree-based classification models have become powerful tools for complex predictive challenges. In this study, the effectiveness of tree-based classification models, such as Random Forest, XGBoost, and Gradient Boosting, is examined on the Titanic survival prediction challenge, which originates from the 1912 Titanic disaster. Passengers survival and death were influenced by various factors in this disaster. By using features such as gender, age, and class, and the survival outcome as the target variable, a binary classification model is developed to predict each passenger's survival status. The study includes data preprocessing, feature selection based on a foundational model, and model training. After the construction, hyper-parameters tuning, and cross-validation of three classifiers, this research compares and analyzes the performance scores to evaluate the characteristics of these tree-based learning methods, aiming to provide a reference for the similar applications.

Topological embedding and directional feature importance in ensemble classifiers for multi-class classification

Cancer is the second leading cause of disease-related death worldwide, and machine learning-based identification of novel biomarkers is crucial for improving early detection and treatment of various cancers. A key challenge in applying machine learning to high-dimensional data is deriving important features in an interpretable manner to provide meaningful insights into the underlying biological mechanismsWe developed a class-based directional feature importance (CLIFI) metric for decision tree methods and demonstrated its use for the The Cancer Genome Atlas proteomics data. The CLIFI metric was incorporated into four algorithms, Random Forest (RF), LAtent VAriable Stochastic Ensemble of Trees (LAVASET), and Gradient Boosted Decision Trees (GBDTs), and a new extension incorporating the LAVA step into GBDTs (LAVABOOST). Both LAVA methods incorporate topological information from protein interactions into the decision function.The different models' performance in classifying 28 cancers resulted in F1-scores of 92.6% (RF), 92.0% (LAVASET), 89.3% (LAVABOOST) and 85.7% (GBDT), with no method outperforming all others for individual cancer type prediction. The CLIFI metric enables visualisation of the model's decision-making functions. The resulting CLIFI value distributions indicated heterogeneity in the expression of several proteins (MYH11, ERα, BCL2) across different cancer types (including brain glioma, breast, kidney, thyroid and prostate cancer) aligning with the original raw expression data.In conclusion, we have developed an integrated, directional feature importance metric for multi-class decision tree-based classification models that facilitates interpretable feature importance assessment. The CLIFI metric can be combined with incorporating topological information into the decision functions of models to introduce inductive bias, enhancing interpretability.

A semiparametric accelerated failure time-based mixture cure tree

The mixture cure rate model (MCM) is the most widely used model for the analysis of survival data with a cured subgroup. In this context, the most common strategy to model the cure probability is to assume a generalized linear model with a known link function, such as the logit link function. However, the logit model can only capture simple effects of covariates on the cure probability. In this article, we propose a new MCM where the cure probability is modeled using a decision tree-based classifier and the survival distribution of the uncured is modeled using an accelerated failure time structure. To estimate the model parameters, we develop an expectation maximization algorithm. Our simulation study shows that the proposed model performs better in capturing nonlinear classification boundaries when compared to the logit-based MCM and the spline-based MCM. This results in more accurate and precise estimates of the cured probabilities, which in-turn results in improved predictive accuracy of cure. We further show that capturing nonlinear classification boundary also improves the estimation results corresponding to the survival distribution of the uncured subjects. Finally, we apply our proposed model and the EM algorithm to analyze an existing bone marrow transplant data.

Tree-based classification model for Long-COVID infection prediction with age stratification using data from the National COVID Cohort Collaborative.

We propose and validate a domain knowledge-driven classification model for diagnosing post-acute sequelae of SARS-CoV-2 infection (PASC), also known as Long COVID, using Electronic Health Records (EHRs) data. We developed a robust model that incorporates features strongly indicative of PASC or associated with the severity of COVID-19 symptoms as identified in our literature review. The XGBoost tree-based architecture was chosen for its ability to handle class-imbalanced data and its potential for high interpretability. Using the training data provided by the Long COVID Computation Challenge (L3C), which was a sample of the National COVID Cohort Collaborative (N3C), our models were fine-tuned and calibrated to optimize Area Under the Receiver Operating characteristic curve (AUROC) and the F1 score, following best practices for the class-imbalanced N3C data. Our age-stratified classification model demonstrated strong performance with an average 5-fold cross-validated AUROC of 0.844 and F1 score of 0.539 across the young adult, mid-aged, and older-aged populations in the training data. In an independent testing dataset, which was made available after the challenge was over, we achieved an overall AUROC score of 0.814 and F1 score of 0.545. The results demonstrated the utility of knowledge-driven feature engineering in a sparse EHR data and demographic stratification in model development to diagnose a complex and heterogeneously presenting condition like PASC. The model's architecture, mirroring natural clinician decision-making processes, contributed to its robustness and interpretability, which are crucial for clinical translatability. Further, the model's generalizability was evaluated over a new cross-sectional data as provided in the later stages of the L3C challenge. The study proposed and validated the effectiveness of age-stratified, tree-based classification models to diagnose PASC. Our approach highlights the potential of machine learning in addressing the diagnostic challenges posed by the heterogeneity of Long-COVID symptoms.

Smart Physiotherapy: Advancing Arm-Based Exercise Classification with PoseNet and Ensemble Models.

Telephysiotherapy has emerged as a vital solution for delivering remote healthcare, particularly in response to global challenges such as the COVID-19 pandemic. This study seeks to enhance telephysiotherapy by developing a system capable of accurately classifying physiotherapeutic exercises using PoseNet, a state-of-the-art pose estimation model. A dataset was collected from 49 participants (35 males, 14 females) performing seven distinct exercises, with twelve anatomical landmarks then extracted using the Google MediaPipe library. Each landmark was represented by four features, which were used for classification. The core challenge addressed in this research involves ensuring accurate and real-time exercise classification across diverse body morphologies and exercise types. Several tree-based classifiers, including Random Forest, Extra Tree Classifier, XGBoost, LightGBM, and Hist Gradient Boosting, were employed. Furthermore, two novel ensemble models called RandomLightHist Fusion and StackedXLightRF are proposed to enhance classification accuracy. The RandomLightHist Fusion model achieved superior accuracy of 99.6%, demonstrating the system's robustness and effectiveness. This innovation offers a practical solution for providing real-time feedback in telephysiotherapy, with potential to improve patient outcomes through accurate monitoring and assessment of exercise performance.

Customer churn analysis using feature optimization methods and tree-based classifiers

Customer churn analysis using feature optimization methods and tree-based classifiers

Optimized Intrusion Detection for IoMT Networks with Tree-Based Machine Learning and Filter-Based Feature Selection.

The Internet of Medical Things (IoMTs) is a network of connected medical equipment such as pacemakers, prosthetics, and smartwatches. Utilizing the IoMT-based system, a huge amount of data is generated, offering experts a valuable resource for tasks such as prediction, real-time monitoring, and diagnosis. To do so, the patient's health data must be transferred to database storage for processing because of the limitations of the storage and computation capabilities of IoMT devices. Consequently, concerns regarding security and privacy can arise due to the limited control over the transmitted information and reliance on wireless transmission, which leaves the network vulnerable to several kinds of attacks. Motivated by this, in this study, we aim to build and improve an efficient intrusion detection system (IDS) for IoMT networks. The proposed IDS leverages tree-based machine learning classifiers combined with filter-based feature selection techniques to enhance detection accuracy and efficiency. The proposed model is used for monitoring and identifying unauthorized or malicious activities within medical devices and networks. To optimize performance and minimize computation costs, we utilize Mutual Information (MI) and XGBoost as filter-based feature selection methods. Then, to reduce the number of the chosen features selected, we apply a mathematical set (intersection) to extract the common features. The proposed method can detect intruders while data are being transferred, allowing for the accurate and efficient analysis of healthcare data at the network's edge. The system's performance is assessed using the CICIDS2017 dataset. We evaluate the proposed model in terms of accuracy, F1 score, recall, precision, true positive rate, and false positive rate. The proposed model achieves 98.79% accuracy and a low false alarm rate 0.007 FAR on the CICIDS2017 dataset according to the experimental results. While this study focuses on binary classification for intrusion detection, we are planning to build a multi-classification approach for future work which will be able to not only detect the attacks but also categorize them. Additionally, we will consider using our proposed feature selection technique for different ML classifiers and evaluate the model's performance empirically in real-world IoMT scenarios.

Condition monitoring of a CNC hobbing cutter using machine learning approach

The state of cutting tools profoundly influences the efficiency of the machining processes within the manufacturing industry. Cutting tool faults are highly undesirable and can adversely impact the performance of machine tools, leading to a shortened operational lifespan. Consequently, it is imperative to minimize power consumption by closely monitoring the condition of cutting tools. This necessitates implementing an effective supervision system to continually assess and predict potential faults. In simpler terms, this entails identifying issues that could compromise the lifespan of cutting tools before they escalate into problems like wear, breakage, or complete failure. This proactive approach ensures the optimal and efficient utilization of cutting tools, reduces the need for maintenance and repair, and enhances process stability, among other benefits. In this paper, a novel approach of machine learning for the condition monitoring of hobbing cutters used in Computer Numerical Control Machines (CNCs) was built. Vibration signals from hobbing blades were recorded under various conditions, including healthy and faulty states. The histogram presents a comprehensive overview of the statistical distribution for five variables related to the studied dataset. MATLAB code and scripts are utilized for extracting relevant statistical features, and for identification of the most relevant features, decision tree algorithms were used. For training the ML algorithms the hyperparameters were selected by the Grid Search Method and the Principal component analysis (PCA) was enabled for the reduction of dimensionality and to simplify the data set. The various conditions of hobbing cutters were then classified using tree-based classification models, giving 100% classification accuracy. It helps to develop a novel condition monitoring system for CNC hobbing cutters using machine learning methods to identify problems in hobbing blades. This would ultimately lead to lower power consumption and enhanced performance of machine tools.

VEXP: A virtual enhanced cross screen panel for off-target pharmacology alerts

We describe the development of the GSK vEXP (virtual enhanced cross screen panel) for off-target pharmacology alerts. The derivation of a panel of machine learning classification models or QSAR models (Quantitative Structure-Activity Relationship) for off-target safety assessment allows early alerting to risk factors in candidate drugs. The models are matched to an internal in-vitro biochemical screening panel described previously with some updates reported here. The extreme imbalance of some internal GSK datasets and most of the related external ChEMBL datasets is shown when considering potency thresholds relevant to in-vitro screening. The small size and bias to the active class make many ChEMBL datasets un-modellable using such thresholds. Although larger, many GSK datasets remain too imbalanced to give a performant model. The value of merging internal and external data to help rebalance datasets and improve the domain of applicability is demonstrated with improvements in model performance frequently seen on merged data. Efforts to collate public datasets with a far better balance of the missing in-actives would likely do more to improve opensource models than simply increasing dataset size. We investigate the use of moving the probability threshold and applying imbalanced learners to help overcome the imbalance problem. Both methods can produce models with improved performance when applied to imbalanced datasets. Datasets with class imbalance 95:5 % or with <100 compounds were un-modellable. Where datasets had a class imbalance of 90:10 % the imbalanced learn methods were often more performant than standard tree-based classifiers. No one classification algorithm consistently out-performed all others and our approach emphasises a standardised, automated build and evaluate approach across all classifiers to identify the best model. The application of vEXP includes ranking of hit compounds for fast prioritisation, flagging of hit series that contain systematic scaffold or functional group related risks and the confirmation that late-stage optimisation is not introducing new off-target activities in established chemical series.

Hyperspectral image construction in different spectral bands of tea leafs for identifying the tea type using O-ConvNet-RF model

Tea, a commonly consumed beverage, is susceptible to being sold in adulterated or expired forms by third-party vendors. Hyperspectral imaging across different wavelength bands has proven to precisely assess the diverse types of tea and their corresponding financial gains. This study aims to employ a deep learning methodology in conjunction with hyperspectral imaging for efficiently classifying tea leaves. A novel approach is proposed, wherein a waveband convolutional neural network is utilized to generate hyper spectral images of tea leaf samples with enhanced resolution. The model known as optimized-convolutional neural network-random forest O- (ConvNet-RF) demonstrated exceptional performance, achieving high accuracy, impressive recall, F1 score, and notable sensitivity rate, outperforming existing alternative methods. The tea leaf types, namely green, yellow, and black, were accurately identified using a combination of the random forest (RF) model and the O-ConvNet-RF model. The tree-based classification method for the identification of tea leaves demonstrated superior performance as compared to alternative machine learning models. In general, this study presents a successful methodology for the classification of tea leaves, with potential implications for consumer processing and distributor profit analysis.

Anomaly-based error and intrusion detection in tabular data: No DNN outperforms tree-based classifiers

Recent years have seen a growing involvement of researchers and practitioners in crafting Deep Neural Networks (DNNs) that seem to outperform existing machine learning approaches for solving classification problems as anomaly-based error and intrusion detection. Undoubtedly, classifiers may be very diverse among themselves, and choosing one or another is typically due to the specific task and target system. Designing and training the optimal tabular data classifier requires extensive experimentation, sensitivity analyses, big datasets, and domain-specific knowledge that may not be available at will or considered a non-strategical asset by many companies and stakeholders. This paper compares, using a total of 23 public datasets: i) traditional (tree-based, statistical) supervised classifiers, ii) DNNs that are specifically designed for classifying tabular data, iii) DNNs for image classification that are applied to tabular data after converting data points into images, alone and as ensembles. Experimental results and related discussions show clear advantages in adopting tree-based classifiers for anomaly-based error and intrusion detection in tabular data as they outperform their competitors, including DNNs. Then, individual classifiers are compared against ensembles using different combinations of the classifiers considered in this study as base-learners, providing a unified final response through many meta-learning strategies. Results show that there is no benefit in building ensembles instead of using a tree-based classifier as Random Forests, eXtreme Gradient Boosting or Extra Trees. The paper concludes that anomaly-based error and intrusion detectors for critical systems should use the old (but gold) tree-based classifiers, which are also easier to fine-tune, and understand; plus, they require less time and resources to learn their model.

PHACTboost: A Phylogeny-Aware Pathogenicity Predictor for Missense Mutations via Boosting.

Most algorithms that are used to predict the effects of variants rely on evolutionary conservation. However, a majority of such techniques compute evolutionary conservation by solely using the alignment of multiple sequences while overlooking the evolutionary context of substitution events. We had introduced PHACT, a scoring-based pathogenicity predictor for missense mutations that can leverage phylogenetic trees, in our previous study. By building on this foundation, we now propose PHACTboost, a gradient boosting tree-based classifier that combines PHACT scores with information from multiple sequence alignments, phylogenetic trees, and ancestral reconstruction. By learning from data, PHACTboost outperforms PHACT. Furthermore, the results of comprehensive experiments on carefully constructed sets of variants demonstrated that PHACTboost can outperform 40 prevalent pathogenicity predictors reported in the dbNSFP, including conventional tools, metapredictors, and deep learning-based approaches as well as more recent tools such as AlphaMissense, EVE, and CPT-1. The superiority of PHACTboost over these methods was particularly evident in case of hard variants for which different pathogenicity predictors offered conflicting results. We provide predictions of 215 million amino acid alterations over 20,191 proteins. PHACTboost is available at https://github.com/CompGenomeLab/PHACTboost. PHACTboost can improve our understanding of genetic diseases and facilitate more accurate diagnoses.

Meta-2OM: A multi-classifier meta-model for the accurate prediction of RNA 2'-O-methylation sites in human RNA.

2'-O-methylation (2-OM or Nm) is a widespread RNA modification observed in various RNA types like tRNA, mRNA, rRNA, miRNA, piRNA, and snRNA, which plays a crucial role in several biological functional mechanisms and innate immunity. To comprehend its modification mechanisms and potential epigenetic regulation, it is necessary to accurately identify 2-OM sites. However, biological experiments can be tedious, time-consuming, and expensive. Furthermore, currently available computational methods face challenges due to inadequate datasets and limited classification capabilities. To address these challenges, we proposed Meta-2OM, a cutting-edge predictor that can accurately identify 2-OM sites in human RNA. In brief, we applied a meta-learning approach that considered eight conventional machine learning algorithms, including tree-based classifiers and decision boundary-based classifiers, and eighteen different feature encoding algorithms that cover physicochemical, compositional, position-specific and natural language processing information. The predicted probabilities of 2-OM sites from the baseline models are then combined and trained using logistic regression to generate the final prediction. Consequently, Meta-2OM achieved excellent performance in both 5-fold cross-validation training and independent testing, outperforming all existing state-of-the-art methods. Specifically, on the independent test set, Meta-2OM achieved an overall accuracy of 0.870, sensitivity of 0.836, specificity of 0.904, and Matthew's correlation coefficient of 0.743. To facilitate its use, a user-friendly web server and standalone program have been developed and freely available at http://kurata35.bio.kyutech.ac.jp/Meta-2OM and https://github.com/kuratahiroyuki/Meta-2OM.

Using Machine Learning for Non-Invasive Detection of Kidney Stones Based on Laboratory Test Results: A Case Study from a Saudi Arabian Hospital.

Kidney stone disease is a widespread urological disorder affecting millions globally. Timely diagnosis is crucial to avoid severe complications. Traditionally, renal stones are detected using computed tomography (CT), which, despite its effectiveness, is costly, resource-intensive, exposes patients to unnecessary radiation, and often results in delays due to radiology report wait times. This study presents a novel approach leveraging machine learning to detect renal stones early using routine laboratory test results. We utilized an extensive dataset comprising 2156 patient records from a Saudi Arabian hospital, featuring 15 attributes with challenges such as missing data and class imbalance. We evaluated various machine learning algorithms and imputation methods, including single and multiple imputations, as well as oversampling and undersampling techniques. Our results demonstrate that ensemble tree-based classifiers, specifically random forest (RF) and extra tree classifiers (ETree), outperform others with remarkable accuracy rates of 99%, recall rates of 98%, and F1 scores of 99% for RF, and 92% for ETree. This study underscores the potential of non-invasive, cost-effective laboratory tests for renal stone detection, promoting prompt and improved medical support.

Distance mapping overlap complexity metric for class-imbalance problems

Data complexity for class-imbalance problems is a hot topic in the field of data mining. Classical data complexity measures use techniques such as kNN, which computes the nearest neighbors of each instance. However, the kNN-based method to obtain the nearest neighbors of all instances is an NP-hard problem, which is not conducive to large-scale data complexity calculations. In order to solve this problem, from the global and local perspectives, four distance map complexity metric methods are proposed, and the complexity values is called the distance map overlap index (DMOI). Firstly, we use Mahalanobis distance or standardized Euclidean distance to calculate the distance of the instance to the center point of the data set, and arrange by the distance value of each instance. Then, according to the sequential label vector, iteratively searches the number of map cut-points for each class. Finally, according to the number of map cut-points of different classes, the DMOI of the data set is calculated. Experiments on 50 class-imbalanced datasets show that the DMOI method outperforms state-of-the-art complexity measures for class-imbalanced problems. Although there is still a gap between DMOI and ONB for rule-based or tree-based classifiers, the use of Pearson correlation coefficient shows that DMOI can effectively approximate ONB, and there is a strong positive correlation between them.

SPE-SHAP: Self-paced ensemble with Shapley additive explanation for the analysis of aviation turbulence triggered by wind shear events

Aviation safety is critically challenged by significant turbulence, often triggered by wind shear events, which jeopardize aircraft controllability and may cause structural damage to the fuselage. Current predictive models, primarily based on Pilot Reports (PIREPs), suffer from data imbalance, undermining their reliability and accuracy. This research addresses the need for a robust approach to accurately estimate and classify aviation turbulence events. We propose a novel framework, the Self-Paced Ensemble (SPE), integrated with Shapley Additive Explanation (SHAP) analysis. This framework utilizes tree-based classifiers such as Gradient Boosting Decision Trees (GBDT), Random Forests (RF), Extreme Gradient Boosting (XGBoost), and Extra Trees (ET) to enhance predictive accuracy and reliability. The SPE model, particularly when combined with XGBoost, demonstrated superior performance with a recall of 70.16 %, specificity of 74.75 %, G-Mean of 73.67 %, Matthews correlation coefficient (MCC) of 0.352, and area under the Receiver Operating Characteristic (ROC) curve of 0.726. Comparatively, the SPE model with RF achieved a recall of 70.05 %, specificity of 73.89 %, G-Mean of 72.57 %, MCC of 0.347, and AU-ROC of 0.721. Post-hoc Wilcoxon signed-rank tests further validated these results. SHAP analysis identified wind shear altitude, magnitude, and causes as the most influential factors affecting significant aviation turbulence prediction. The SPE-SHAP framework not only significantly improves the prediction of aviation turbulence but also enhances model transparency by elucidating key contributing factors. These insights can inform updates to aircraft operation protocols and enhance flight training, thereby increasing safety during wind shear events.