Feature Engineering Techniques Research Articles

Advanced Trans-BiGRU-QA Fusion Model for Atmospheric Mercury Prediction

With the deepening of the Industrial Revolution and the rapid development of the chemical industry, the large-scale emissions of corrosive dust and gases from numerous factories have become a significant source of air pollution. Mercury in the atmosphere, identified by the United Nations Environment Programme (UNEP) as one of the globally concerning air pollutants, has been proven to pose a threat to the human environment with potential carcinogenic risks. Therefore, accurately predicting atmospheric mercury concentration is of critical importance. This study proposes a novel advanced model—the Trans-BiGRU-QA hybrid—designed to predict the atmospheric mercury concentration accurately. Methodology includes feature engineering techniques to extract relevant features and applies a sliding window technique for time series data preprocessing. Furthermore, the proposed Trans-BiGRU-QA model is compared to other deep learning models, such as GRU, LSTM, RNN, Transformer, BiGRU, and Trans-BiGRU. This study utilizes air quality data from Vietnam to train and test the models, evaluating their performance in predicting atmospheric mercury concentration. The results show that the Trans-BiGRU-QA model performed exceptionally well in terms of Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and R-squared (R2), demonstrating high accuracy and robustness. Compared to other deep learning models, the Trans-BiGRU-QA model exhibited significant advantages, indicating its broad potential for application in environmental pollution prediction.

Multidisciplinary Framework for Vibration-Based Gear Fault Diagnosis Using Experiments, Modeling, and Machine Learning

Vibration-based gear diagnosis is crucial for ensuring the reliability of rotating machinery, making the monitoring of gear health essential for preventing costly downtime and optimizing performance. This study proposes a multidisciplinary framework to enhance gear diagnosis, that aligns with the new era of digital twins by integrating experiments, dynamic modeling, physical preprocessing, and machine learning. Within this framework, we focus on three core procedures: domain adaptation to reduce discrepancies between measured data and synthetic data generated by dynamic models; physical preprocessing, grounded in in-depth investigations dictating signal processing and feature engineering techniques; and learning algorithms, encompassing the process of training AI-based models. We demonstrate this framework through a comprehensive case study of localized tooth fault diagnosis, using controlled-degradation tests and realistic simulations. First, we detect faults using unsupervised learning algorithms; then, we use zero-shot-learning for classifying between localized and distributed faults; finally, we adopt a one-shot-learning strategy for severity estimation. Above all, this hybrid framework bridges the gaps between physical-based and AI-based approaches by combining physical knowledge and advanced algorithmics with machine learning. This contributes to the PHM field by offering valuable insights into integrating different aspects of research, thereby enhancing performance in gear diagnosis tasks.

Identification of a Novel Biomarker Panel for Breast Cancer Screening.

Breast cancer remains a major public health concern, and early detection is crucial for improving survival rates. Metabolomics offers the potential to develop non-invasive screening and diagnostic tools based on metabolic biomarkers. However, the inherent complexity of metabolomic datasets and the high dimensionality of biomarkers complicates the identification of diagnostically relevant features, with multiple studies demonstrating limited consensus on the specific metabolites involved. Unlike previous studies that rely on singular feature selection techniques such as Partial Least Square (PLS) or LASSO regression, this research combines supervised and unsupervised machine learning methods with random sampling strategies, offering a more robust and interpretable approach to feature selection. This study aimed to identify a parsimonious and robust set of biomarkers for breast cancer diagnosis using metabolomics data. Plasma samples from 185 breast cancer patients and 53 controls (from the Cooperative Human Tissue Network, USA) were analyzed. This study also overcomes the common issue of dataset imbalance by using propensity score matching (PSM), which ensures reliable comparisons between cancer and control groups. We employed Univariate Naïve Bayes, L2-regularized Support Vector Classifier (SVC), Principal Component Analysis (PCA), and feature engineering techniques to refine and select the most informative features. Our best-performing feature set comprised 11 biomarkers, including 9 metabolites (SM(OH) C22:2, SM C18:0, C0, C3OH, C14:2OH, C16:2OH, LysoPC a C18:1, PC aa C36:0 and Asparagine), a metabolite ratio (Kynurenine-to-Tryptophan), and 1 demographic variable (Age), achieving an area under the ROC curve (AUC) of 98%. These results demonstrate the potential for a robust, cost-effective, and non-invasive breast cancer screening and diagnostic tool, offering significant clinical value for early detection and personalized patient management.

Forecasting Patient Early Readmission from Irish Hospital Discharge Records Using Conventional Machine Learning Models.

Predicting patient readmission is an important task for healthcare risk management, as it can help prevent adverse events, reduce costs, and improve patient outcomes. In this paper, we compare various conventional machine learning models and deep learning models on a multimodal dataset of electronic discharge records from an Irish acute hospital. We evaluate the effectiveness of several widely used machine learning models that leverage patient demographics, historical hospitalization records, and clinical diagnosis codes to forecast future clinical risks. Our work focuses on addressing two key challenges in the medical fields, data imbalance and the variety of data types, in order to boost the performance of machine learning algorithms. Furthermore, we also employ SHapley Additive Explanations (SHAP) value visualization to interpret the model predictions and identify both the key data features and disease codes associated with readmission risks, identifying a specific set of diagnosis codes that are significant predictors of readmission within 30 days. Through extensive benchmarking and the application of a variety of feature engineering techniques, we successfully improved the area under the curve (AUROC) score from 0.628 to 0.7 across our models on the test dataset. We also revealed that specific diagnoses, including cancer, COPD, and certain social factors, are significant predictors of 30-day readmission risk. Conversely, bacterial carrier status appeared to have minimal impact due to lower case frequencies. Our study demonstrates how we effectively utilize routinely collected hospital data to forecast patient readmission through the use of conventional machine learning while applying explainable AI techniques to explore the correlation between data features and patient readmission rate.

Data-centric automated approach to predict autism spectrum disorder based on selective features and explainable artificial intelligence.

Autism spectrum disorder (ASD) is a neurodevelopmental condition marked by notable challenges in cognitive function, understanding language, recognizing objects, interacting with others, and communicating effectively. Its origins are mainly genetic, and identifying it early and intervening promptly can reduce the necessity for extensive medical treatments and lengthy diagnostic procedures for those impacted by ASD. This research is designed with two types of experimentation for ASD analysis. In the first set of experiments, authors utilized three feature engineering techniques (Chi-square, backward feature elimination, and PCA) with multiple machine learning models for autism presence prediction in toddlers. The proposed XGBoost 2.0 obtained 99% accuracy, F1 score, and recall with 98% precision with chi-square significant features. In the second scenario, main focus shifts to identifying tailored educational methods for children with ASD through the assessment of their behavioral, verbal, and physical responses. Again, the proposed approach performs well with 99% accuracy, F1 score, recall, and precision. In this research, cross-validation technique is also implemented to check the stability of the proposed model along with the comparison of previously published research works to show the significance of the proposed model. This study aims to develop personalized educational strategies for individuals with ASD using machine learning techniques to meet their specific needs better.

Current Status and Challenges and Future Trends of Deep Learning-Based Intrusion Detection Models.

With the advancement of deep learning (DL) technology, DL-based intrusion detection models have emerged as a focal point of research within the domain of cybersecurity. This paper provides an overview of the datasets frequently utilized in the research. This article presents an overview of the widely utilized datasets in the research, establishing a basis for future investigation and analysis. The text subsequently summarizes the prevalent data preprocessing methods and feature engineering techniques utilized in intrusion detection. Following this, it provides a review of seven deep learning-based intrusion detection models, namely, deep autoencoders, deep belief networks, deep neural networks, convolutional neural networks, recurrent neural networks, generative adversarial networks, and transformers. Each model is examined from various dimensions, highlighting their unique architectures and applications within the context of cybersecurity. Furthermore, this paper broadens its scope to include intrusion detection techniques facilitated by the following two large-scale predictive models: the BERT series and the GPT series. These models, leveraging the power of transformers and attention mechanisms, have demonstrated remarkable capabilities in understanding and processing sequential data. In light of these findings, this paper concludes with a prospective outlook on future research directions. Four key areas have been identified for further research. By addressing these issues and advancing research in the aforementioned areas, this paper envisions a future in which DL-based intrusion detection systems are not only more accurate and efficient but also better aligned with the dynamic and evolving landscape of cybersecurity threats.

Detecting the Severity of Depression in Online Forum Data by Leveraging Implicit Semantic Inferences

Depression, a prevalent mental health disorder with global implications, exerts a profound negative influence on individuals’ lives. While the prediction of depression (as a binary classification task) is a well-established research area, depression severity detection is a new research direction with limited studies. In the context of detecting the severity of depression through online forum data, this research endeavors to offer two distinct solutions by employing Ada embeddings, GPT 3.5 Turbo, and LIWC as feature engineering techniques, while AutoSklearn serves as the ensemble learning algorithm. Notably, the outcomes of this study significantly outperform existing state-of-the-art models on both depression severity annotated datasets used in this research. The results also showcase the potential reuse nature of the proposed models in diverse data sources due to their high performance in both datasets. Furthermore, as a valuable practical outcome, a software prototype has been developed, capable of providing the depression severity level, along with associated symptoms and keywords, upon inputting an online forum post.

Enhancing Large-Diameter Tunnel Construction Safety with Robust Optimization and Machine Learning Integrated into BIM

Aim This study aims to enhance safety in large diameter tunnel construction by integrating robust optimization and machine learning (ML) techniques with Building Information Modeling (BIM). By acquiring and preprocessing various datasets, implementing feature engineering, and using algorithms like SVM, decision trees, ANN, and random forests, the study demonstrates the effectiveness of ML models in risk prediction and mitigation, ultimately advancing safety performance in civil engineering projects. Background Large diameter tunnel construction presents significant safety challenges. Traditional methods often fall short of effectively predicting and mitigating risks. This study addresses these gaps by integrating robust optimization and machine learning (ML) approaches with Building Information Modeling (BIM) technology. By acquiring and preprocessing diverse datasets, implementing feature engineering, and employing ML algorithms, the study aims to enhance risk prediction and safety measures in tunnel construction projects. Objective The objective of this study is to improve safety in large diameter tunnel construction by integrating robust optimization and machine learning (ML) techniques with Building Information Modeling (BIM). This involves acquiring and preprocessing diverse datasets, using feature engineering to extract key parameters, and applying ML algorithms like SVM, decision trees, ANN, and random forests to predict and mitigate risks, ultimately enhancing safety performance in civil engineering projects. Methods The study's methods include acquiring and preprocessing various datasets (geological, structural, environmental, operational, historical, and simulation). Feature engineering techniques are used to extract key safety parameters for tunnels. Machine learning algorithms, such as decision trees, support vector machines (SVM), artificial neural networks, and random forests, are employed to analyze the data and predict construction risks. The SVM algorithm, with a 98.76% accuracy, is the most reliable predictor. Results The study found that the Support Vector Machine (SVM) algorithm was the most accurate predictor of risks in large diameter tunnel construction, achieving a 98.76% accuracy rate. Other models, such as decision trees, artificial neural networks, and random forests, also performed well, validating the effectiveness of ML-based solutions for risk assessment and mitigation. These predictive models enable stakeholders to monitor construction, allocate resources, and implement preventative measures effectively. Conclusion The study concludes that integrating machine learning (ML) approaches with Building Information Modeling (BIM) significantly improves safety in large diameter tunnel construction. The Support Vector Machine (SVM) algorithm, with 98.76% accuracy, is the most reliable predictor of risks. Other models, like decision trees, artificial neural networks, and random forests, also perform well, validating ML-based solutions for risk assessment. Adopting these ML approaches enhances safety performance and resource management in civil engineering projects.

Household electricity consumption prediction using database combinations, ensemble and hybrid modeling techniques

Household electricity consumption (HEC) is changing over time, depends on multiple factors, and leads to effects on the prediction accuracy of the model. The objective of this work is to propose a novel methodology for improving HEC prediction accuracy. This study uses two original datasets, namely questionnaire survey (QS) and monthly consumption (MC), which contain data from 225 consumers from Maharashtra, India. The original datasets are combined to create three additional datasets, namely QS + MC, QS equation (QsEq) + next month’s consumptions, and QsEq + MC. Furthermore, the HEC prediction accuracy is boosted by applying different approaches, like correlation methods, feature engineering techniques, data quality assessment, heterogeneous ensemble prediction (HEP), and the hybrid model. Five HEP models are created using dataset combinations and machine learning algorithms. Based on the MC dataset, the random forest provides the best prediction of RMSE (36.18 kWh), MAE (25.73 kWh), and R2 (0.76). Similarly, QsEq + MC dataset adaptive boosting provides a better prediction of RMSE (36.77 kWh), MAE (26.18 kWh), and R2 (0.76). This prediction accuracy is further increased using the proposed hybrid model to RMSE (22.02 kWh), MAE (13.04 kWh), and R2 (0.92). This research work benefits researchers, policymakers, and utility companies in obtaining accurate prediction models and understanding HEC.

Multi-iteration active learning for the composition design of potassium–sodium niobate ceramics with enhanced piezoelectric coefficient

The piezoelectricity of piezoceramics considerably depends on the doped compositions. However, the conventional trial-and-error approach to composition design is time-consuming and inefficient when searching for high-performance piezoceramics with various dopants. In this study, we introduce a data-driven framework to accelerate the design and discovery of lead-free (K,Na)NbO3 (KNN) materials with enhanced piezoelectric performance. The proposed framework integrates machine learning with feature engineering, surrogate-based optimisation, and experimentation in an active learning loop. Using feature engineering techniques, we demonstrated that the piezoelectric coefficient d33 of KNN materials can be predicted based on a set of elemental and atomic properties. We evaluated six machine learning models and found that support vector regression with a radial basis function kernel achieved high accuracy in predicting the d33 values of KNN ceramics. After three iterations of experimentation and optimisation, we identified an optimal composition with a relatively high d33 of ∼353 pC/N from thousands of possible compositions using a pure exploitation strategy. This study demonstrates an efficient and systematic approach to the composition design of KNN-based piezoceramics, which can be applied to investigate the diverse functional properties of other materials.

Research on wind turbine icing prediction data processing and accuracy of machine learning algorithm

Studying the icing problem of wind turbine blades is crucial for optimizing wind farm operation and maintenance. Traditional machine learning algorithms like Random Forest and Support Vector Machines have limitations in handling complex time series data and capturing long-term dependencies, leading to insufficient accuracy and generalization. To address these gaps, this paper proposes a comprehensive approach involving advanced machine learning frameworks and feature engineering techniques. This paper based on the original SACDA dataset, four distinct types of processing are performed on the prediction data via PCA dimensionality reduction and the introduction of new features; meanwhile, the four types mentioned above of datasets are trained and predicted using the Gated Recycling Unit model (GRU), Random Forest (RF), GA-BP Neural Network (BP), and Extreme Learning Machine (ELM) models. The results indicate that the prediction accuracies of all four types of models are more satisfactory, with the RF model having the highest prediction accuracy and the overall accuracy remaining above 99 percent, the dataset + RF model after dimensionality reduction of the original sensitive features having the highest prediction accuracy and speed.

Hyperparameter Optimization for Convolutional Neural Network-Based Sentiment Analysis

The rapid development of digital technology brings about the importance of product or service opinions as a data source in business intelligence. Sentiment analysis is a business intelligence tool that can help gauge market trends by analysing the available opinions. Various studies have been developed to solve sentiment analysis problems, from selecting feature engineering techniques to selecting sentiment classification models based on classical or deep learning methods. However, these studies still select hyperparameters on a trial basis, which poses problems in terms of time and not optimal performance. Therefore, this research proposes the implementation of Bayesian Optimization to automate the selection of hyperparameters in sentiment analysis based on a Convolutional Neural Network (CNN). The results show that implementing the Bayesian Optimization method has the lowest computation time compared to Random and Grid Searches, which is only 175.746765 seconds, in the fifth trial. In addition, it made a reasonably large cut in terms of times compared to manual trial-based hyperparameter tuning that needs 540 trials. In the model assessment process, implementing Bayesian Optimization gives the highest values on Precision and Specificity, which are 0.8168 and 0.7176, respectively. Thus, implementing Bayesian Optimization is the right choice for sentiment analysis tasks since the precision of predicting negative sentiment classes is vital in hospitality business intelligence, especially the use of related information for product improvement or hotel service improvement. The implementation of Bayesian Optimization not only applied as a reliable hyperparameter selection technique for classification tasks but also regression prediction tasks.

Predicting Pseudomonas aeruginosa drug resistance using artificial intelligence and clinical MALDI-TOF mass spectra.

Matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS) is widely used in clinical microbiology laboratories for bacterial identification but its use for detection of antimicrobial resistance (AMR) remains limited. Here, we used MALDI-TOF MS with artificial intelligence (AI) approaches to successfully predict AMR in Pseudomonas aeruginosa, a priority pathogen with complex AMR mechanisms. The highest performance was achieved for modern β-lactam/β-lactamase inhibitor drugs, namely, ceftazidime/avibactam and ceftolozane/tazobactam. For these drugs, the model demonstrated area under the receiver operating characteristic curve (AUROC) of 0.869 and 0.856, specificity of 0.925 and 0.897, and sensitivity of 0.731 and 0.714, respectively. As part of this work, we developed dynamic binning, a feature engineering technique that effectively reduces the high-dimensional feature set and has wide-ranging applicability to MALDI-TOF MS data. Compared to conventional feature engineering approaches, the dynamic binning method yielded highest performance in 7 of 10 antimicrobials. Moreover, we showcased the efficacy of transfer learning in enhancing the AUROC performance for 8 of 11 antimicrobials. By assessing the contribution of features to the model's prediction, we identified proteins that may contribute to AMR mechanisms. Our findings demonstrate the potential of combining AI with MALDI-TOF MS as a rapid AMR diagnostic tool for Pseudomonas aeruginosa.IMPORTANCEPseudomonas aeruginosa is a key bacterial pathogen that causes significant global morbidity and mortality. Antimicrobial resistance (AMR) emerges rapidly in P. aeruginosa and is driven by complex mechanisms. Drug-resistant P. aeruginosa is a major challenge in clinical settings due to limited treatment options. Early detection of AMR can guide antibiotic choices, improve patient outcomes, and avoid unnecessary antibiotic use. Matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS) is widely used for rapid species identification in clinical microbiology. In this study, we repurposed mass spectra generated by MALDI-TOF and used them as inputs for artificial intelligence approaches to successfully predict AMR in P. aeruginosa for multiple key antibiotic classes. This work represents an important advance toward using MALDI-TOF as a rapid AMR diagnostic for P. aeruginosa in clinical settings.

Systematic Literature Review on Data Preprocessing for Improved Water Potability Prediction: A Study of Data Cleaning, Feature Engineering, and Dimensionality Reduction Techniques

Systematic Literature Review on Data Preprocessing for Improved Water Potability Prediction: A Study of Data Cleaning, Feature Engineering, and Dimensionality Reduction Techniques

A multi-dimensional machine learning framework for accurate and efficient battery state of charge estimation

Accurate state of charge (SOC) estimation is essential for battery safe and efficient utilization. As artificial intelligence technologies evolve, data-driven methods have become mainstream for estimating SOC. However, the technique can significantly deteriorate model performance when encountering poor or insufficient data quality. In this paper, we apply median filtering to eliminate extreme noise and utilize continuous wavelet transform to extract time-frequency features from voltage signals. Additionally, we generate novel features via feature crossing. We then apply dimensionality reduction via the random forest method to decrease computational expense. Finally, we select a convolutional neural network (CNN) as the base model to learn optimized features for more precise SOC estimation. To confirm the efficacy of our proposed method, this study compares it with CNN, long short-term memory (LSTM), bidirectional LSTM (BILSTM), and a CNN-BILSTM model combined with an attention mechanism. These comparisons are conducted under different temperatures and operating conditions. The results indicate that this method achieves a mean absolute error and a root mean square error of less than 2.89 % and 3.71 %, respectively, in SOC estimation, demonstrating superior accuracy compared to other models. This study underscores the significance of feature engineering techniques in SOC estimation.

Temporal Characterization and Visualization of Revolving Therapy-Events in Lung Cancer Patients.

This paper presents a comprehensive workflow for integrating revolving events into the transitive sequential pattern mining (tSPM+) algorithm and Machine Learning for Health Outcomes (MLHO) framework, emphasizing best practices and pitfalls in its application. We emphasize feature engineering and visualization techniques, demonstrating their efficacy in capturing temporal relationships. Applied to an EGFR lung cancer cohort, our approach showcases reliable temporal insights even in a small dataset. This work highlights the importance of temporal nuances in healthcare data analysis, paving the way for improved disease understanding and patient care.

A Machine Learning Framework for Melt-Pool Geometry Prediction and Process Parameter Optimization in the Laser Powder-Bed Fusion Process

Abstract This study presents a cost-effective and high-precision machine learning (ML) method for predicting the melt-pool geometry and optimizing the process parameters in the laser powder-bed fusion (LPBF) process with Ti-6Al-4V alloy. Unlike many ML models, the presented method incorporates five key features, including three process parameters (laser power, scanning speed, and spot size) and two material parameters (layer thickness and powder porosity). The target variables are the melt-pool width and depth that collectively define the melt-pool geometry and give insight into the melt-pool dynamics in LPBF. The dataset integrates information from an extensive literature survey, computational fluid dynamics (CFD) modeling, and laser melting experiments. Multiple ML regression methods are assessed to determine the best model to predict the melt-pool geometry. Tenfold cross-validation is applied to evaluate the model performance using five evaluation metrics. Several data pre-processing, augmentation, and feature engineering techniques are performed to improve the accuracy of the models. Results show that the “Extra Trees regression” and “Gaussian process regression” models yield the least errors for predicting melt-pool width and depth, respectively. The ML modeling results are compared with the experimental and CFD modeling results to validate the proposed ML models. The most influential parameter affecting the melt-pool geometry is also determined by the sensitivity analysis. The processing parameters are optimized using an iterative grid search method employing the trained ML models. The presented ML framework offers computational speed and simplicity, which can be implemented in other additive manufacturing techniques to comprehend the critical traits.

Exploring the predictive factors of heart disease using rare association rule mining

Cardiovascular diseases continue to be the leading cause of mortality worldwide, claiming a significant number of lives each year. Despite the advancements in predictive models, including logistic regression, neural networks, and random forests, these techniques often lack transparency and interpretability, limiting their practical application in clinical settings. To address this challenge, this research introduces EPFHD-RARMING, an innovative approach designed to enhance the understanding and predictability of heart disease through the discovery of rare and meaningful patterns. EPFHD-RARMING utilizes rare association rule mining to uncover hidden and unexpected rules that identify critical factors contributing to heart disease. This method is particularly adept at identifying high-risk patterns in individuals who appear healthy but may develop heart disease under certain conditions, thus facilitating early intervention and preventive measures. By integrating these insights with established feature engineering techniques, EPFHD-RARMING enhances its practical utility, enabling medical professionals to proactively manage patient care and tailor interventions to individual risk profiles. This study demonstrates the effectiveness of EPFHD-RARMING in providing a deeper, actionable understanding of the complex dynamics of heart disease. The model’s ability to identify and interpret rare patterns holds significant promise for advancing medical analytics and improving patient outcomes. Moreover, the applicability of EPFHD-RARMING extends beyond the healthcare domain, offering valuable insights in various fields where the discovery of rare patterns is critical, such as finance, marketing, and cybersecurity. This study conducts a comprehensive evaluation, which demonstrates the superior performance of EPFHD-RARMING compared to traditional predictive models in identifying key factors contributing to heart disease, in terms of interestingness, explainability, and comprehensiveness of insights. The results underscore the potential of this innovative approach to revolutionize our understanding and prediction of heart disease, ultimately contributing to more effective and personalized healthcare solutions. This research emphasizes the importance of rare association rule mining in medical analytics and paves the way for future studies to explore and utilize these techniques across diverse domains.

Applications and issues of artificial intelligence in the financial sector

This paper investigates the application of artificial intelligence in the financial sector, analyzing the existing technical, ethical, and legal issues, and proposing corresponding solutions. The research background highlights the widespread use of AI technology in the financial industry and the efficiency and cost benefits it brings. The research focuses on challenges related to data quality, feature engineering, model complexity, real-time capability, computational resources, and data privacy protection. The research method includes literature review and case analysis, revealing the applications of AI technology in stock prediction, risk management, trading strategy optimization, and customer service. The results indicate that effective data cleaning, automated feature engineering, model simplification and regularization techniques, the use of interpretability tools like LIME and SHAP, and the introduction of fairness evaluation standards can significantly enhance AI model performance and transparency. The conclusion points out that these measures can not only solve the current technical and ethical issues of AI in the financial sector but also promote the widespread application and standardization of AI technology in other fields.

Integrating Feature Selection with Machine Learning for Accurate Reservoir Landslide Displacement Prediction

Accurate prediction of reservoir landslide displacements is crucial for early warning and hazard prevention. Current machine learning (ML) paradigms for predicting landslide displacement demonstrate superior performance, while often relying on various feature engineering techniques, such as decomposing into different temporal lags and feature selection. This study investigates the impact of various feature selection techniques on the performance of ML algorithms for landslide displacement prediction. The Shuping and Baishuihe landslides in China’s Three Gorges Reservoir Area are used to comprehensively benchmark four prevalent ML algorithms. Both static ML models, including backpropagation neural network (BPNN), support vector machine (SVM), and dynamic models, such as long short-term memory (LSTM), and gated recurrent unit (GRU), are included. Each ML model is evaluated under three feature engineering techniques: raw multivariate time series, and feature selection under maximal information coefficient-partial autocorrelation function (MIC-PACF), or grey relational analysis-PACF (GRA-PACF). The results demonstrate that appropriate feature selection methods could significantly improve the performance of static ML models. In contrast, dynamic models effectively leverage inherent capabilities in capturing temporal dynamics within raw multivariate time series, seeing marginal gains with extensive feature engineering compared to no feature selection strategy. The optimal feature selection approach varies based on the ML model and specific landslide, highlighting the importance of case-specific assessments. The findings in this study offer guidance on integrating feature selection techniques with different machine learning models to maximize the robustness and generalizability of data-driven landslide displacement prediction frameworks.