predictive modeling

To address the low prediction accuracy and AUC in diabetes classification on high-dimensional, low-sample-size (HDLSS) NHANES health data, this paper proposes an XGBoost-MLP hybrid model. It first employs XGBoost for robust feature encoding and importance-based dimensionality reduction, yielding compact, dense low-dimensional representations; these are then fed into a lightweight multilayer perceptron (MLP) for nonlinear classification. This architecture synergistically integrates the interpretability and feature selection capability of tree-based models with the strong nonlinear modeling capacity of neural networks. Evaluated on a rigorously preprocessed NHANES subset, the model achieves an AUC of 0.892 and a balanced accuracy of 0.831—significantly outperforming logistic regression, random forest, and standalone XGBoost baselines (p < 0.01). The implementation code and full reproducibility scripts are publicly available.

Current evaluation practices for supervised learning models are often misleading due to an overreliance on single aggregate metrics, which neglect the alignment among data characteristics, task objectives, and real-world application contexts. This work reframes model evaluation as a context-dependent, decision-oriented process and systematically investigates—through controlled experiments—the impact of dataset properties, validation strategies, class imbalance, and asymmetric error costs on evaluation outcomes. Leveraging diverse benchmark datasets, multiple validation protocols, and multidimensional performance measures, the study uncovers common pitfalls such as the accuracy paradox, data leakage, and metric misuse. It proposes a structured evaluation framework explicitly aligned with operational goals, offering principled guidance for developing more robust, reliable, and trustworthy supervised learning systems.

This study addresses the low efficiency of incident classification in aviation safety investigations. We establish the first systematic evaluation framework to assess five mainstream supervised learning models—SVM, logistic regression, random forest, XGBoost, and KNN—on a binary classification task distinguishing “incidents” from “serious incidents.” Our empirical analysis reveals, for the first time, that SMOTE significantly degrades classification performance, indicating its unsuitability for this domain. Across 100 repeated experiments, random forest achieves superior performance (accuracy: 0.77; F1-score: 0.78; Matthews correlation coefficient: 0.51). We further develop and deploy an interactive web application, now integrated into operational safety analysis workflows. The core contributions are: (1) a robust, empirically validated model selection paradigm for aviation safety text classification; and (2) domain-specific data preprocessing guidelines, notably advising against SMOTE use in this context.

This paper addresses the fundamental trade-off between zero false negatives and low false positives in dynamic classification. To resolve this, we propose a lightweight multi-model collaborative framework. Methodologically, we introduce a novel self-supervised classification learning mechanism; dynamically partition input data into $N$ mutually exclusive subsets; train independent submodels for parallel prediction; and incorporate a confidence-threshold-based filtering and prediction rejection mechanism—eliminating unreliable predictions without requiring auxiliary verification models. Supervised feedback is further leveraged to iteratively refine model performance. Experiments demonstrate strict zero false negatives and a 37.2% reduction in false positive rate over state-of-the-art ensemble methods under low partitioning error; under high partitioning error, the framework maintains robustness comparable to current best models. Our core contribution is a reliability- and efficiency-aware lightweight paradigm for dynamic classification.

To address the challenge of accurately predicting delivery lead times in automotive flexible manufacturing zones—characterized by non-paced production and highly variable work orders—this paper proposes a context-aware supervised classification framework, circumventing the limitations of conventional continuous regression models when applied to discrete, long-tailed delivery time distributions. Methodologically, it introduces, for the first time in non-cyclic production lines, a systematic evaluation of lightweight gradient-boosting models (specifically LightGBM), augmented by a periodic dynamic retraining mechanism to accommodate line evolution. Feature engineering employs one-hot encoding, while model selection and hyperparameter optimization involve comparative evaluation across LightGBM, XGBoost, CatBoost, and SVM. Experimental results demonstrate that the optimal LightGBM-based classifier achieves a relative prediction accuracy of 90%, substantially outperforming existing non-AI systems. This provides a reliable, production-deployable metric for scheduling decisions and delivery commitment management.

This work addresses the rigidity in model selection and poor interpretability inherent in conventional XGBoost–neural network ensembles. We propose an adaptive fusion framework grounded in dynamic meta-learning, which jointly leverages uncertainty quantification and feature importance as dual control signals to guide fine-grained scheduling and weighted integration of the two base models at inference time. Our key innovation lies in co-modeling uncertainty estimates and interpretability-aware metrics—specifically, feature importance—within the meta-learner’s decision process, thereby simultaneously enhancing predictive performance and decision transparency. Extensive experiments across multiple benchmark datasets demonstrate that our method consistently outperforms static ensembles and individual baselines, achieving average accuracy gains of 2.1–4.7 percentage points. Moreover, it provides auditable, instance-level rationale for model selection and feature-level attribution, supporting both reliability assessment and human-understandable explanations.

This study addresses the challenge of accurately forecasting daily log returns of the Nepal Stock Exchange (NEPSE) index in a high-noise, nonlinear emerging market context. To this end, it proposes a time series prediction framework based on XGBoost that integrates lagged returns and technical indicators—such as the Relative Strength Index (RSI) and rolling volatility. The methodology employs an expanding-window forward-rolling validation scheme combined with temporal cross-validation to mitigate look-ahead bias, alongside Optuna for hyperparameter optimization. Empirical results demonstrate that the optimal configuration, featuring 20 lags and an expanding window, achieves a test-set RMSE of 0.013450, MAE of 0.009814, and directional accuracy of 65.15%, significantly outperforming ARIMA and Ridge regression baselines. This work thus offers a reproducible and robust solution for financial time series forecasting in emerging markets.

This work challenges the prevailing paradigm that time-series representation learning requires task-specific pretraining, investigating whether pretrained time-series forecasting models can serve as general-purpose feature extractors for time-series classification. Method: We evaluate zero-shot transfer of frozen forecasting models to classification tasks and propose two model-agnostic embedding enhancement strategies to facilitate cross-task representation reuse. Contribution/Results: Empirical results across multiple classification benchmarks show that the best-performing forecasting models achieve classification accuracy competitive with or superior to dedicated classification-pretrained models. Moreover, forecasting capability exhibits a strong positive correlation with downstream classification performance. This study establishes forecasting as an effective proxy task for learning transferable, efficient, and generalizable time-series representations—paving a new pathway toward universal time-series foundation models.

This study addresses the challenges of severe class imbalance and temporal uncertainty in multivariate time-series anomaly detection for steam turbines. We systematically evaluate ensemble and hybrid approaches, proposing a lightweight segmentation-based ensemble model that integrates change-point detection, clustering-based substructure representation, and Random Forest/XGBoost—without relying on complex feature engineering or hybrid architectures. Experimental results demonstrate that this streamlined approach significantly outperforms sophisticated methods in robustness, interpretability, and deployment efficiency: it achieves an AUC-ROC of 0.976, an F1-score of 0.41, and guarantees 100% early anomaly detection within the prescribed time window. Our key contribution lies in empirically establishing that, in real-world industrial settings, jointly optimizing segmentation strategy and model simplicity yields greater practical value than architectural complexity alone.

This study addresses the challenge classical frequentist statisticians face in understanding neural networks by proposing a reconstruction of neural networks through the lens of linear regression. By simplifying network architecture and integrating statistical interpretability techniques, the approach reformulates deep learning models into a modeling paradigm familiar to statisticians. The method preserves the expressive power of neural networks while offering intuitive parameter interpretations and customizable pathways, thereby significantly lowering the cognitive barrier for statisticians entering the field of deep learning. The resulting framework balances theoretical rigor with practical usability, fostering meaningful integration and methodological exchange between traditional statistics and modern deep learning.