arXiv:2511.13952v1 Announce Type: new 
Abstract: Random Forests (RFs) typically train each tree on a bootstrap sample of the same size as the training set, i.e., bootstrap rate (BR) equals 1.0. We systematically examine how varying BR from 0.2 to 5.0 affects RF performance across 39 heterogeneous regression datasets and 16 RF configurations, evaluating with repeated two-fold cross-validation and mean squared error. Our results demonstrate that tuning the BR can yield significant improvements over the default: the best setup relied on BR \leq 1.0 for 24 datasets, BR > 1.0 for 15, and BR = 1.0 was optimal in 4 cases only. We establish a link between dataset characteristics and the preferred BR: datasets with strong global feature-target relationships favor higher BRs, while those with higher local target variance benefit from lower BRs. To further investigate this relationship, we conducted experiments on synthetic datasets with controlled noise levels. These experiments reproduce the observed bias-variance trade-off: in low-noise scenarios, higher BRs effectively reduce model bias, whereas in high-noise settings, lower BRs help reduce model variance. Overall, BR is an influential hyperparameter that should be tuned to optimize RF regression models.

تدرس هذه الدراسة تأثير تغيير معدل أخذ العينات بالأسلوب البوتستراب (BR) على أداء الغابات العشوائية (RF) في مهام الانحدار. تقليديًا، تستخدم RF معدل BR يساوي 1.0، حيث يتم تدريب كل شجرة على عينة بوتستراب تعادل حجم مجموعة التدريب. تحلل هذه البحث 39 مجموعة بيانات انحدار متنوعة و16 تكوينًا لـ RF، باستخدام التحقق المتقاطع ذو الطيّتين المتكرر وخطأ المربعات المتوسطة للتقييم. تشير النتائج إلى أن ضبط معدل BR يمكن أن يؤدي إلى تحسينات كبيرة، مع اختلاف القيم المثلى بناءً على خصائص مجموعة البيانات.

El estudio investiga el efecto de variar la tasa de muestreo bootstrap (BR) en el rendimiento de los Bosques Aleatorios (RF) en tareas de regresión. Tradicionalmente, los RF utilizan una BR de 1.0, entrenando cada árbol en una muestra bootstrap igual al tamaño del conjunto de entrenamiento. Esta investigación analiza 39 conjuntos de datos de regresión diversos y 16 configuraciones de RF, utilizando validación cruzada de dos pliegues repetida y error cuadrático medio para la evaluación. Los hallazgos indican que ajustar la BR puede llevar a mejoras significativas, con valores óptimos de BR que …

Cette étude examine l'effet de la variation du taux d'échantillonnage bootstrap (BR) sur la performance des forêts aléatoires (RF) dans les tâches de régression. Traditionnellement, les RF utilisent un BR de 1,0, formant chaque arbre sur un échantillon bootstrap égal à la taille de l'ensemble d'entraînement. Cette recherche analyse 39 ensembles de données de régression divers et 16 configurations de RF, en utilisant une validation croisée à deux volets répétée et l'erreur quadratique moyenne pour l'évaluation. Les résultats indiquent que l'ajustement du BR peut entraîner des améliorations sign…

The study investigates the effect of varying the bootstrap sampling rate (BR) on the performance of Random Forests (RFs) in regression tasks. Traditionally, RFs use a BR of 1.0, training each tree on a bootstrap sample equal to the training set size. This research analyzes 39 diverse regression datasets and 16 RF configurations, employing repeated two-fold cross-validation and mean squared error for evaluation. Findings indicate that adjusting the BR can lead to significant performance enhancements, with optimal BR values varying based on dataset characteristics.

The Impact of Bootstrap Sampling Rate on Random Forest Performance in Regression Tasks

arXiv:2510.12416v2 Announce Type: replace 
Abstract: We introduce a novel high-frequency daily panel dataset of both markets and news-based indicators -- including Geopolitical Risk, Economic Policy Uncertainty, Trade Policy Uncertainty, and Political Sentiment -- for 42 countries across both emerging and developed markets. Using this dataset, we study how sentiment dynamics shape sovereign risk, measured by Credit Default Swap (CDS) spreads, and evaluate their forecasting value relative to traditional drivers such as global monetary policy and market volatility. Our horse-race analysis of forecasting models demonstrates that incorporating news-based indicators significantly enhances predictive accuracy and enriches the analysis, with non-linear machine learning methods -- particularly Random Forests -- delivering the largest gains. Our analysis reveals that while global financial variables remain the dominant drivers of sovereign risk, geopolitical risk and economic policy uncertainty also play a meaningful role. Crucially, their effects are amplified through non-linear interactions with global financial conditions. Finally, we document pronounced regional heterogeneity, as certain asset classes and emerging markets exhibit heightened sensitivity to shocks in policy rates, global financial volatility, and geopolitical risk.

تم تقديم مجموعة بيانات جديدة يومية عالية التردد تشمل الأسواق ومؤشرات قائمة على الأخبار مثل المخاطر الجيوسياسية وعدم اليقين في السياسة الاقتصادية في 42 دولة. تتيح هذه المجموعة تحليل كيفية تأثير ديناميكيات المشاعر على المخاطر السيادية، التي تقاس من خلال فروق مقايضات التخلف عن السداد، وتبرز القوة التنبؤية للمؤشرات القائمة على الأخبار مقارنة بالعوامل الاقتصادية التقليدية.

Se ha introducido un nuevo conjunto de datos de panel diario de alta frecuencia que abarca mercados e indicadores basados en noticias, como el Riesgo Geopolítico y la Incertidumbre de Política Económica en 42 países. Este conjunto de datos permite analizar cómo las dinámicas de sentimiento influyen en el riesgo soberano, medido a través de los spreads de los swaps de incumplimiento crediticio, y destaca el poder predictivo de los indicadores basados en noticias sobre los impulsores económicos tradicionales.

Un nouvel ensemble de données de panel quotidien à haute fréquence a été introduit, englobant les marchés et des indicateurs basés sur les nouvelles tels que le risque géopolitique et l'incertitude de la politique économique dans 42 pays. Cet ensemble de données permet d'analyser comment les dynamiques de sentiment influencent le risque souverain, mesuré par les spreads des swaps de défaut de crédit, et met en évidence le pouvoir prédictif des indicateurs basés sur les nouvelles par rapport aux moteurs économiques traditionnels.

A novel high-frequency daily panel dataset has been introduced, encompassing markets and news-based indicators such as Geopolitical Risk and Economic Policy Uncertainty across 42 countries. This dataset allows for an analysis of how sentiment dynamics influence sovereign risk, measured through Credit Default Swap spreads, and highlights the predictive power of news-based indicators over traditional economic drivers.

Geopolitics, Geoeconomics and Risk: A Machine Learning Approach

arXiv:2512.00397v1 Announce Type: new 
Abstract: Random Forests and Gradient Boosting are among the most effective algorithms for supervised learning on tabular data. Both belong to the class of tree-based ensemble methods, where predictions are obtained by aggregating many randomized regression trees. In this paper, we develop a theoretical framework for analyzing such methods through Reproducing Kernel Hilbert Spaces (RKHSs) constructed on tree ensembles -- more precisely, on the random partitions generated by randomized regression trees. We establish fundamental analytical properties of the resulting Random Forest kernel, including boundedness, continuity, and universality, and show that a Random Forest predictor can be characterized as the unique minimizer of a penalized empirical risk functional in this RKHS, providing a variational interpretation of ensemble learning. We further extend this perspective to the continuous-time formulation of Gradient Boosting introduced by Dombry and Duchamps, and demonstrate that it corresponds to a gradient flow on a Hilbert manifold induced by the Random Forest RKHS. A key feature of this framework is that both the kernel and the RKHS geometry are data-dependent, offering a theoretical explanation for the strong empirical performance of tree-based ensembles. Finally, we illustrate the practical potential of this approach by introducing a kernel principal component analysis built on the Random Forest kernel, which enhances the interpretability of ensemble models, as well as GVI, a new geometric variable importance criterion.

تم تطوير إطار نظري جديد لتحليل طرق التجميع القائمة على الأشجار، وخاصة الغابات العشوائية وزيادة التدرج، من خلال مساحات هيلبرت القابلة للتكرار (RKHS). يوفر هذا الإطار رؤى حول الخصائص التحليلية للغابات العشوائية، بما في ذلك الحدودية والاستمرارية، ويقدم تفسيرًا تباينيًا للتعلم الجماعي.

Se ha desarrollado un nuevo marco teórico para analizar los métodos de conjunto basados en árboles, en particular los Bosques Aleatorios y el Aumento de Gradiente, a través de los Espacios de Hilbert Reproducibles (RKHS). Este marco proporciona información sobre las propiedades analíticas de los Bosques Aleatorios, incluyendo la acotación y la continuidad, y ofrece una interpretación variacional del aprendizaje en conjunto.

Un nouveau cadre théorique a été développé pour analyser les méthodes d'ensemble basées sur des arbres, en particulier les forêts aléatoires et le boosting par gradient, à travers les espaces de Hilbert reproduisants (RKHS). Ce cadre fournit des informations sur les propriétés analytiques des forêts aléatoires, y compris la bornitude et la continuité, et offre une interprétation variationnelle de l'apprentissage par ensemble.

A new theoretical framework has been developed for analyzing tree-based ensemble methods, particularly Random Forests and Gradient Boosting, through Reproducing Kernel Hilbert Spaces (RKHS). This framework provides insights into the analytical properties of Random Forests, including boundedness and continuity, and offers a variational interpretation of ensemble learning.

The Impact of Bootstrap Sampling Rate on Random Forest Performance in Regression Tasks

Was this article worth reading? Share it

Random Generator Pro

FastML

Reviewradar