Feature Importance Evaluation Research Articles

Employing machine learning for enhanced abdominal fat prediction in cavitation post-treatment

This study investigates the application of cavitation in non-invasive abdominal fat reduction and body contouring, a topic of considerable interest in the medical and aesthetic fields. We explore the potential of cavitation to alter abdominal fat composition and delve into the optimization of fat prediction models using advanced hyperparameter optimization techniques, Hyperopt and Optuna. Our objective is to enhance the predictive accuracy of abdominal fat dynamics post-cavitation treatment. Employing a robust dataset with abdominal fat measurements and cavitation treatment parameters, we evaluate the efficacy of our approach through regression analysis. The performance of Hyperopt and Optuna regression models is assessed using metrics such as mean squared error, mean absolute error, and R-squared score. Our results reveal that both models exhibit strong predictive capabilities, with R-squared scores reaching 94.12% and 94.11% for post-treatment visceral fat, and 71.15% and 70.48% for post-treatment subcutaneous fat predictions, respectively. Additionally, we investigate feature selection techniques to pinpoint critical predictors within the fat prediction models. Techniques including F-value selection, mutual information, recursive feature elimination with logistic regression and random forests, variance thresholding, and feature importance evaluation are utilized. The analysis identifies key features such as BMI, waist circumference, and pretreatment fat levels as significant predictors of post-treatment fat outcomes. Our findings underscore the effectiveness of hyperparameter optimization in refining fat prediction models and offer valuable insights for the advancement of non-invasive fat reduction methods. This research holds important implications for both the scientific community and clinical practitioners, paving the way for improved treatment strategies in the realm of body contouring.

A Comprehensive Evaluation of Dual-Polarimetric Sentinel-1 SAR Data for Monitoring Key Phenological Stages of Winter Wheat

Large-scale crop phenology monitoring is critical for agronomic planning and yield prediction applications. Synthetic Aperture Radar (SAR) remote sensing is well-suited for crop growth monitoring due to its nearly all-weather observation capability. Yet, the capability of the dual-polarimetric SAR data for wheat phenology estimation has not been thoroughly investigated. Here, we conducted a comprehensive evaluation of Sentinel-1 SAR polarimetric parameters’ sensibilities on winter wheat’s key phenophases while considering the incidence angle. We extracted 12 polarimetric parameters based on the covariance matrix and a dual-pol-version H-α decomposition. All parameters were evaluated by their temporal profile and feature importance score of Gini impurity with a decremental random forest classification process. A final wheat phenology classification model was built using the best indicator combination. The result shows that the Normalized Shannon Entropy (NSE), Degree of Linear Polarization (DoLP), and Stokes Parameter g2 were the three most important indicators, while the Span, Average Alpha (α2¯), and Backscatter Coefficient σVH0 were the three least important features in discriminating wheat phenology for all three incidence angle groups. The smaller-incidence angle (30–35°) SAR images are better suited for estimating wheat phenology. The combination of NSE, DoLP, and two Stokes Parameters (g2 and g0) constitutes the most effective indicator ensemble. For all eight key phenophases, the average Precision and Recall scores were above 0.8. This study highlighted the potential of dual-polarimetric SAR data for wheat phenology estimation. The feature importance evaluation results provide a reference for future phenology estimation studies using dual-polarimetric SAR data in choosing better-informed indicators.

Precision in Prevention: Tailoring Single-Use Negative Pressure Wound Therapy Utilization Through Artificial Intelligence-Based Surgical Site Complications Risk and Cost Modeling.

Background: Surgical site complications (SSCs) are common, yet preventable hospital-acquired conditions. Single-use negative pressure wound therapy (sNPWT) has been shown to be effective in reducing rates of these complications. In the era of value-based care, strategic allocation of sNPWT is needed to optimize both clinical and financial outcomes. Materials and Methods: We conducted a retrospective analysis using data from the Premier Healthcare Database (2017-2021) for 10 representative open procedures in orthopedic, abdominal, cardiovascular, cesarean delivery, and breast surgery. After separating data into training and validation sets, various machine learning algorithms were used to develop pre-operative SSC risk prediction models. Model performance was assessed using standard metrics and predictors of SSCs were identified through feature importance evaluation. Highest-performing models were used to simulate the cost-effectiveness of sNPWT at both the patient and population level. Results: The prediction models demonstrated good performance, with an average area under the curve of 76%. Prominent predictors across subspecialities included age, obesity, and the level of procedure urgency. Prediction models enabled a simulation analysis to assess the population-level cost-effectiveness of sNPWT, incorporating patient and surgery-specific factors, along with the established efficacy of sNPWT for each surgical procedure. The simulation models uncovered significant variability in sNPWT's cost-effectiveness across different procedural categories. Conclusions: This study demonstrates that machine learning models can effectively predict a patient's risk of SSC and guide strategic utilization of sNPWT. This data-driven approach allows for optimization of clinical and financial outcomes by strategically allocating sNPWT based on personalized risk assessments.

Does Economic Stability Influence Family Development? Insights from Women in Korea with the Lowest Childbirth Rates Worldwide

The aim of this study is to explore the multidimensional relationships among factors influencing decision-making processes regarding women’s willingness to marry and childbirth in South Korea with recognizing the context of family development in East Asian cultures. To this end, we employed three different analytical approaches, including classification tree modeling, Cox proportional hazard modeling, and permutation feature importance evaluation. Leveraging longitudinal data specific to Korean women, we highlighted the significance of socio-economic factors in family development dynamics. Our findings revealed that financial stability played a crucial role. Unmarried women’s willingness to marry was influenced by their perspectives on economic stability, while households’ consumption capacity and financial capability determined childbirth decisions and timing. We observed a trend of postponed marriage among women in their marriageable age range, particularly those with stable economic situations, reflecting a prevalent trend of skepticism of marriage in Korean society. Additional findings related to values, cultural factors, and personal happiness also suggested the challenges that discourage younger generations from entering into marriage and starting families in South Korea. By offering insights into these dynamics, our study provides practical implications for addressing the obstacles faced, contributing to a better understanding of family development dynamics.

Combination of Machine Learning and Raman Spectroscopy for Determination of the Complex of Whey Protein Isolate with Hyaluronic Acid.

Macromolecules and their complexes remain interesting topics in various fields, such as targeted drug delivery and tissue regeneration. The complex chemical structure of such substances can be studied with a combination of Raman spectroscopy and machine learning. The complex of whey protein isolate (WPI) and hyaluronic acid (HA) is beneficial in terms of drug delivery. It provides HA properties with the stability obtained from WPI. However, differences between WPI-HA and WPI solutions can be difficult to detect by Raman spectroscopy. Especially when the low HA (0.1, 0.25, 0.5% w/v) and the constant WPI (5% w/v) concentrations are used. Before applying the machine learning techniques, all the collected data were divided into training and test sets in a ratio of 3:1. The performances of two ensemble methods, random forest (RF) and gradient boosting (GB), were evaluated on the Raman data, depending on the type of problem (regression or classification). The impact of noise reduction using principal component analysis (PCA) on the performance of the two machine learning methods was assessed. This procedure allowed us to reduce the number of features while retaining 95% of the explained variance in the data. Another application of these machine learning methods was to identify the WPI Raman bands that changed the most with the addition of HA. Both the RF and GB could provide feature importance data that could be plotted in conjunction with the actual Raman spectra of the samples. The results show that the addition of HA to WPI led to changes mainly around 1003 cm-1 (correspond to ring breath of phenylalanine) and 1400 cm-1, as demonstrated by the regression and classification models. For selected Raman bands, where the feature importance was greater than 1%, a direct evaluation of the effect of the amount of HA on the Raman intensities was performed but was found not to be informative. Thus, applying the RF or GB estimators to the Raman data with feature importance evaluation could detect and highlight small differences in the spectra of substances that arose from changes in the chemical structure; using PCA to filter out noise in the Raman data could improve the performance of both the RF and GB. The demonstrated results will make it possible to analyze changes in chemical bonds during various processes, for example, conjugation, to study complex mixtures of substances, even with small additions of the components of interest.

CIRI-Deep Enables Single-Cell and Spatial Transcriptomic Analysis of Circular RNAs with Deep Learning.

Circular RNAs (circRNAs) are a crucial yet relatively unexplored class of transcripts known for their tissue- and cell-type-specific expression patterns. Despite the advances in single-cell and spatial transcriptomics, these technologies face difficulties in effectively profiling circRNAs due to inherent limitations in circRNA sequencing efficiency. To address this gap, a deep learning model, CIRI-deep, is presented for comprehensive prediction of circRNA regulation on diverse types of RNA-seq data. CIRI-deep is trained on an extensive dataset of 25 million high-confidence circRNA regulation events and achieved high performances on both test and leave-out data, ensuring its accuracy in inferring differential events from RNA-seq data. It is demonstrated that CIRI-deep and its adapted version enable various circRNA analyses, including cluster- or region-specific circRNA detection, BSJ ratio map visualization, and trans and cis feature importance evaluation. Collectively, CIRI-deep's adaptability extends to all major types of RNA-seq datasets including single-cell and spatial transcriptomic data, which will undoubtedly broaden the horizons of circRNA research.

Partial multi-label learning via robust feature selection and relevance fusion optimization

Partial Multi-Label Learning (PML) is a more practical learning paradigm, in which the labeling information is ambiguated. Most existing PML algorithms rely on assumptions to resolve ambiguity. However, these assumptions do not account for the origin of the noise labeling and therefore fail to address the impact of noise on the learner’s performance at the root. In this paper, we will propose a PML method jointly granular ball-based robust feature selection and relevance fusion optimization (PML-GR). Specifically, in the first stage, we construct a granular ball to compute the core-set with weights and then design a feature importance evaluation function to assign weights to each feature in the core-set, resulting in a ranking of feature importance for the PML learner; in the second stage, based on the selected features, a fusion-based objective function is constructed to compute the label confidence by taking into account the joint effect of the global sample similarity and local label relevance. Finally, a multi-label prediction model is learned by fitting the multi-output regressor to the label confidence. The experimental results demonstrate that the proposed method achieves competitive generalization performance by effective feature selection and relevance fusion optimization, which can focus more on discriminative features and minimize the effect of noisy labels during training.

Development of 1D-convolutional Neural Network-based Height Profile Prediction Model in Directed Energy Deposition Process Using Melt-pool Image Data

Directed energy deposition (DED) process is a representative metal additive manufacturing technology that uses a flexible deposition head mainly used for repairs in space and marine industries. The DED process saves time and money as it repairs only damaged parts and components. Therefore, a geometric control is important to fill the volume of the target damaged area economically and accurately. However, efficiency depends on process parameters such as laser power, scanning speed. This study proposes a one-dimensional convolutional neural network (1D-CNN) model to predict the height profile of the DED parts utilizing melt-pool image data. First, DED experiments were performed for a total of nine cases considering laser power and scanning speed as parameters. The collected melt-pool image data was pre-processed and only those related to the regions of interest were extracted. Initially, a total of 15 features were extracted from size, shape, location, and brightness from the melt-pool images. Then, 10 critical ones, selected through a permutation feature importance evaluation method, were input to the 1D-CNN algorithm to predict height profiles of the deposited layers. In testing phase, a mean absolute percentage error (MAPE) of 9.55% was achieved, and thus, applicability of the proposed model was verified.

Coronary stenosis quantification in cardiac computed tomography angiography: multi-factorial optimization of image quality and radiation dose.

The accuracy and variability of quantification in computed tomography angiography (CTA) are affected by the interplay of imaging parameters and patient attributes. The assessment of these combined effects has been an open engineering challenge. In this study, we developed a framework that optimizes imaging parameters for accurate and consistent coronary stenosis quantification in cardiac CTA while accounting for patient-specific variables. The framework utilizes a task-specific image quality index, the estimability index (), approximated by a surrogate estimability polynomial function (EPF) capable of finding the optimal protocol that (1)maximizes image quality with an upper bound for desired radiation dose or (2)minimizes the dose level with a lower bound of acceptable image quality. The optimization process was formulated with the decision variables being subject to a set of constraints. The methodology was verified using CTA data from a prior clinical trial (prospective multi-center imaging study for evaluation of chest pain) by assessing the concordance of its prediction with the trial results. Further, the framework was used to derive an optimum protocol for each case based on the patient attributes, gauging how much improvement would have been possible if the derived optimized protocol would have been deployed. The framework produced results consistent with imaging physics principles with approximated EPFs of 97% accuracy. The feature importance evaluation demonstrated a close match with earlier studies. The verification study found scores closely predicting the cardiologist scores to within 95% in terms of the area under the receiver operating characteristic curve and predicting potential for either an average of fourfold increase in within a targeted dose or a reduction in radiation dose by an average of 57% without reducing the image quality. The protocol optimization framework provides means to assess and optimize CTA in terms of either image quality or radiation dose objectives with its results predicting prior clinical trial findings.

Using Natural Language Processing to Identify Stigmatizing Language in Labor and Birth Clinical Notes.

Stigma and bias related to race and other minoritized statuses may underlie disparities in pregnancy and birth outcomes. One emerging method to identify bias is the study of stigmatizing language in the electronic health record. The objective of our study was to develop automated natural language processing (NLP) methods to identify two types of stigmatizing language: marginalizing language and its complement, power/privilege language, accurately and automatically in labor and birth notes. We analyzed notes for all birthing people > 20 weeks' gestation admitted for labor and birth at two hospitals during 2017. We then employed text preprocessing techniques, specifically using TF-IDF values as inputs, and tested machine learning classification algorithms to identify stigmatizing and power/privilege language in clinical notes. The algorithms assessed included Decision Trees, Random Forest, and Support Vector Machines. Additionally, we applied a feature importance evaluation method (InfoGain) to discern words that are highly correlated with these language categories. For marginalizing language, Decision Trees yielded the best classification with an F-score of 0.73. For power/privilege language, Support Vector Machines performed optimally, achieving an F-score of 0.91. These results demonstrate the effectiveness of the selected machine learning methods in classifying language categories in clinical notes. We identified well-performing machine learning methods to automatically detect stigmatizing language in clinical notes. To our knowledge, this is the first study to use NLP performance metrics to evaluate the performance of machine learning methods in discerning stigmatizing language. Future studies should delve deeper into refining and evaluating NLP methods, incorporating the latest algorithms rooted in deep learning.

A hybrid modelling approach for detecting seasonal variations in inland Green-Blue Ecosystems

Deforestation, environmental pollution, and the overexploitation of resources, in addition to the Earth's natural cycles, are scaling up the impacts of climate change in the provision of Ecosystem Services (ES). Green-Blue Ecosystems (GBE) are impacted by climatic conditions, topography, and water presence. Data-driven modelling techniques may effectively capture the effects of seasonal variations while modelling natural ecosystems. This research proposes a hybrid modelling approach that combines Deep Learning and traditional Machine Learning, Sensitivity Analysis and Feature Importance Evaluation (FIE) to investigate seasonality effects on mapping GBE. The models, built using satellite imagery from the Spring and Summer seasons of the Mediterranean climate zone, included spectral indices, topography (DEM), and groundwater depth (GD). The model that best suited the analysis was selected using sensitivity tests and hyperparameter optimization. The study shows that land cover classes of transitional woodland shrubs, inland marshes, cultivated land parcels, and watercourses are better classified in the Spring, with an accuracy of 0.814. FIE indicates that spectral indices are the most important predictors for detecting green ecosystems in both seasons. Additionally, DEM and GD are the most relevant predictors to classify watercourses in the Summer. An analytical examination of the input data and hyperparameter settings facilitates understanding of models' behaviour while improving models' prediction. This research provides an advanced understanding of the effects of seasonal variations on the status of GBE and enhances understanding of modelling ES in areas with a growing need for changes in land use and high water supply demand.

PDHF: Effective phishing detection model combining optimal artificial and automatic deep features

Currently, the increasing number of high-volume phishing attacks is among the largest threats to networking environments on a daily basis. During such a severe attack, researchers prefer to extract numerous features to improve the accuracy of phishing detection. However, the redundant features that may exist in the extracted feature set may be adapted by phishing attackers, not only degrading detection performance but also shortening the effective time of the constructed detection models. To address these problems, this study proposes phishing detection based on hybrid features (PDHF), a novel phishing detection model based on a combination of optimal artificial and automatic deep learning features. The optimal artificial phishing features are obtained by removing redundant features based on the newly designed feature importance evaluation index and an improved bidirectional search algorithm. To extend the effective time of phishing detection, deep features are learned from URLs using a one-dimensional character convolutional neural network (CNN) and a disorderly quantized attention mechanism. The experimental results show that PDHF outperforms many state-of-the-art methods and achieves an accuracy of 0.9965, precision of 0.9942, recall of 0.9940, and F1-score of 0.9941. These results can help in the development of a security plug-in for clients, browsers, and various instant messaging tools that run on network edges, personal computers, smartphones, and other personal terminals.

Development and initial evaluation of the Hearing Aid Attribute and Feature Importance Evaluation (HAFIE) questionnaire

Objectives To develop and validate a novel questionnaire aimed at providing a structured, evidence-based methodology for hearing aid recommendation and selection using self-reported importance ratings for different modern hearing aid features. Design The initial questionnaire items were created through a concept mapping approach that involved input from hearing aid users. Hearing care professional focus groups (n = 10) were conducted to assess questionnaire content and design, and to guide modifications. Validation of this initial 34-item version of the questionnaire was conducted using an anonymous online survey tool (Qualtrics). Exploratory factor analysis was used to assess the factor structure of the dataset, using principal axis factoring. Questionnaire reliability and inter-item correlation were assessed. Items with low factor loading and high cross-loading were removed. Study sample Two hundred and eighteen adult participants with a self-reported hearing loss (median age = 48 years, range = 18–95 years) completed the questionnaire. Results Analysis and item removal resulted in a 28-item questionnaire. Three factors were identified, dividing the hearing aid features into the subscales: “Advanced connectivity & streaming”, “Physical attributes & usability”, and “Sound quality & intelligibility”. Conclusion This study has resulted in a patient-oriented questionnaire that allows clinicians to gather patient input in a structured manner.

Machine learning models for predicting compressive strength of fiber-reinforced concrete containing waste rubber and recycled aggregate

The compressive strength of fiber-reinforced rubberized recycled aggregate concrete (FR3C) is an important performance indicator for its practical application and durability in the concrete industry. The constituents of this concrete exhibit distinct mechanical properties, particularly the waste rubber and fiber components. The mix design for desired compressive strength of such concrete is critical and depends on the constituents' internal complex relationships. Therefore, this study proposes machine learning-based models to predict the compressive strength and internal dependencies among the constituents of the FR3C's. Given this, a dataset consisting of 905 experimental data of various mix proportions were compiled to train and test thirteen different machine learning models, namely linear regression, ridge regression, lasso regression, support vector machine, k-nearest neighbors, artificial neural network, decision tree, random forest, AdaBoost, Voting Regressor, Gradient Boost, CatBoost, and XGBoost. The input characteristics were water/cement ratio (W/C), percentage of rubber, replacement level of recycled concrete aggregate (RCA), percentage of fiber, and its type. The results suggest that the CatBoost is the most accurate model for predicting the compressive strength of FR3C, with the highest coefficient of determination, lowest root mean squared error (RMSE), and low mean absolute error on the test data. Feature importance evaluation showed that w/c ratio, nominal aggregate size, and rubber (%) are important critical parameters that significantly influence the compressive strength of FR3C.

Explainability meets uncertainty quantification: Insights from feature-based model fusion on multimodal time series

Feature importance evaluation is one of the prevalent approaches to interpreting Machine Learning (ML) models. A drawback of using these methods for high-dimensional datasets is that they often lead to high-dimensional explanation output that hinders human analysis. This is especially true for explaining multimodal ML models, where the problem’s complexity is further exacerbated by the inclusion of multiple data modalities and an increase in the overall number of features. This work proposes a novel approach to lower the complexity of feature-based explanations. The proposed approach is based on uncertainty quantification techniques, allowing for a principled way of reducing the number of modalities required to explain the model’s predictions. We evaluated our method in three multimodal datasets comprising physiological time series. Results show that the proposed method can reduce the complexity of the explanations while maintaining a high level of accuracy in the predictions. This study illustrates an innovative example of the intersection between the disciplines of uncertainty quantification and explainable artificial intelligence.

Prediction model of atrial fibrillation recurrence after Cox-Maze IV procedure in patients with chronic valvular disease and atrial fibrillation based on machine learning algorithm.

Atrial fibrillation (AF) is a prevalent cardiac arrhythmia, and Cox-maze IV procedure (CMP-IV) is a commonly employed surgical technique for its treatment. Currently, the risk factors for atrial fibrillation recurrence following CMP-IV remain relatively unclear. In recent years, machine learning algorithms have demonstrated immense potential in enhancing diagnostic accuracy, predicting patient outcomes, and devising personalized treatment strategies. This study aims to evaluate the efficacy of CMP-IV on treating chronic valvular disease with AF, utilize machine learning algorithms to identify potential risk factors for AF recurrence, construct a CMP-IV postoperative AF recurrence prediction model. A total of 555 patients with AF combined with chronic valvular disease, who met the criteria, were enrolled from January 2012 to December 2019 from the Second Xiangya Hospital of Central South University and the Affiliated Xinqiao Hospital of the Army Medical University, with an average age of (57.95±7.96) years, including an AF recurrence group (n=117) and an AF non-recurrence group (n=438). Kaplan-Meier method was used to analyze the sinus rhythm maintenance rate, and 9 machine learning models were developed including random forest, gradient boosting decision tree (GBDT), extreme gradient boosting (XGBoost), bootstrap aggregating, logistic regression, categorical boosting (CatBoost), support vector machine, adaptive boosting, and multi-layer perceptron. Five-fold cross-validation and model evaluation indicators [including F1 score, accuracy, precision, recall, and area under the curve (AUC)] were used to evaluate the performance of the models. The 2 best-performing models were selected for further analyze, including feature importance evaluation and Shapley additive explanations (SHAP) analysis, identifying AF recurrence risk factors, and building an AF recurrence risk prediction model. The 5-year sinus rhythm maintenance rate for the patients was 82.13% (95% CI 78.51% to 85.93%). Among the 9 machine learning models, XGBoost and CatBoost models performed best, with the AUC of 0.768 (95% CI 0.742 to 0.786) and 0.762 (95% CI 0.723 to 0.801), respectively. Feature importance and SHAP analysis showed that duration of AF, preoperative left ventricular ejection fraction, postoperative heart rhythm, preoperative neutrophil-to-lymphocyte ratio, preoperative left atrial diameter, preoperative heart rate, and preoperative white blood cell were important factors for AF recurrence. Conclusion: Machine learning algorithms can be effectively used to identify potential risk factors for AF recurrence after CMP-IV. This study successfuly constructs 2 prediction model which may enhance individualized treatment plans.

CellBRF: a feature selection method for single-cell clustering using cell balance and random forest.

Single-cell RNA sequencing (scRNA-seq) offers a powerful tool to dissect the complexity of biological tissues through cell sub-population identification in combination with clustering approaches. Feature selection is a critical step for improving the accuracy and interpretability of single-cell clustering. Existing feature selection methods underutilize the discriminatory potential of genes across distinct cell types. We hypothesize that incorporating such information could further boost the performance of single cell clustering. We develop CellBRF, a feature selection method that considers genes' relevance to cell types for single-cell clustering. The key idea is to identify genes that are most important for discriminating cell types through random forests guided by predicted cell labels. Moreover, it proposes a class balancing strategy to mitigate the impact of unbalanced cell type distributions on feature importance evaluation. We benchmark CellBRF on 33 scRNA-seq datasets representing diverse biological scenarios and demonstrate that it substantially outperforms state-of-the-art feature selection methods in terms of clustering accuracy and cell neighborhood consistency. Furthermore, we demonstrate the outstanding performance of our selected features through three case studies on cell differentiation stage identification, non-malignant cell subtype identification, and rare cell identification. CellBRF provides a new and effective tool to boost single-cell clustering accuracy. All source codes of CellBRF are freely available at https://github.com/xuyp-csu/CellBRF.

An integrated approach based on the correction of imbalanced small datasets and the application of machine learning algorithms to predict total phosphorus concentration in rivers

Increased concentrations of Total Phosphorus (TP) in freshwater systems lead to eutrophication and can contribute to a wide range of environmental effects. In the modern era, water quality models have increasingly been used globally for the development of management scenarios with the aim of reducing the eutrophication risk. However, the accuracy of these models is limited by the quality of the boundary conditions forcing data, namely TP concentration datasets. In this study, a novel methodology is proposed to improve machine learning prediction accuracy in the modeling of river TP concentration forced with small input training datasets. These models can then be used to increase the quality and consistency of the TP concentration datasets required to force water quality models. This new methodology relies on the generation of 100 new training datasets from the raw training datasets of input predictors through the implementation of an over/undersampling technique. The modeling approach used in this study was supported by the application of ten machine learning algorithms to estimate the TP concentration values in 22 rivers located in Portugal. The modeling approach also included an input feature importance evaluation, as well as model hyperparameter optimization. In general terms, the Extreme Gradient Boosting (XGBoost) and Support Vector Regressor (SVR) models performed best overall, with the ensemble results recorded for both models working to increase the mean Nash-Sutcliffe efficiency (NSE) across all the areas being studied by 96% (0.01 ± 0.22 to 0.31 ± 0.32) and reduce the mean percentage bias (PBIAS) by 43% (18.47 ± 17.31 to 10.60 ± 17.40). The results of this study suggest that the solution proposed has the potential to significantly improve the modeling of TP concentration in rivers with machine learning methods, as well as providing increased scope for its application to larger training datasets and the prediction of other types of dependent variables. Hopefully, the results of this study will further add to the body of information available in this area of research and aid the development of the water management process.

Land Use/Land Cover Change Analysis Using Multi-Temporal Remote Sensing Data: A Case Study of Tigris and Euphrates Rivers Basin

Multi-temporal land use/land cover (LULC) change analysis is essential for environmental planning and recourses management. Various global LULC datasets are available now. However, they do not show strong consistency on a regional scale and are mainly time limited. Therefore, high-quality multi-temporal LULC mapping with reasonable consistency on a regional scale is still demanding. In this study, using the Landsat 7, Landsat 8, and the NASA digital elevation model (DEM), LULC mapping of the Tigris and Euphrates rivers basin (TEB) was performed by random forest (RF) classifier in the Google Earth Engine platform during 2000–2022. The spectral bands, spectral indices, morphological, and textural features were applied in the developed procedure. The results indicated that the proposed approach had accurate performance (accuracy = 0.893 and an F score = 0.820) with a good consistency with previous studies. The feature importance evaluation was carried out using Gini index, and spectral indices were identified as the most important features in LULC mapping. Overall, severe LULC change has happened in the TEB during the last two decades. Our results revealed the expansion of water and built-up classes while trees class has experienced a decreasing trend. From a regional perspective, three main areas in the east and south-east of Iraq, north-west of Iraq, and east of Syria were identified where LULC change was intense. These areas are prone to land degradation and dust storms emission problems, and it is necessary to take steps to prevent severe LULC changes in them.

Rethinking the regularity in mobility patterns of personal vehicle drivers: A multi‐city comparison using a feature engineering approach

AbstractLarge‐scale trajectory data offer a finer lens into the regularity in individual mobility choices. Previous studies have exerted efforts to measure the regularity in people's location visiting patterns. However, the complexity of travel behavior at different spatial and temporal scales has not been adequately considered. To capture regularity in a more comprehensive manner, we construct human mobility profiles with interpretable features at three levels, that is, location, motif, and route, on personal vehicle drivers. A feature engineering approach is designed to analyze the extent to which individuals exhibit multi‐level regularity. The analysis pipeline includes feature selection, user segmentation and profiling, and feature importance evaluation. Our empirical study analyzed over 4 million trips of 3743 personal vehicle drivers collected over a month in six metropolitan areas in the United States. The weak correlations between features confirm the validity of quantifying regularity from different aspects. We discovered five clusters of drivers (i.e., gig drivers, homebodies, movers, typical drivers, and work‐focused commuters) that differ in their regularity to commute to the workplace and the inclination to participate in non‐work activities. A similar driver segmentation and profiling pattern is found in all of the studied metro areas. The minor differences are interpreted from the distribution of mobility features and urban features. The proposed method using multi‐level feature engineering provides a generic framework to study regularity and can be readily adapted to other mobility data sources by customizing the features. The improved understanding of mobility patterns within the built environment is valuable for innovating urban transportation solutions.