The article examines the impact of inflationary spikes in the early 2020s on the adequacy of actuarial reserves in the property and casualty (P&C) segment, identifying the drivers of systematic delays in adjusting these reserves. The relevance of the study is determined by the fact that loss reserves constitute the core of an insurer’s financial resilience. At the same time, the structural break in inflation trends from 2021 to 2024 disrupted the baseline stationarity premise upon which classical reserving methodologies implicitly rest. The objective is to provide an integrated assessment of how macro-inflationary shocks transmit into claim severity and generate the phenomenon of inflation recognition lag, as well as to compare the sensitivity of traditional approaches (Chain Ladder, Bornhuetter–Ferguson) to the new inflation regime. Scientific novelty lies in integrating a macro-level analysis of CPI and industry-specific inflation drivers with a micro-level examination of Allstate’s public reporting (10-K) and a quantitative assessment of the temporal lag in reserve responses to inflation. It is shown that parameter inertia in Chain Ladder and the anchoring of  Bornhuetter–Ferguson to a priori expectations create an under-reserving risk in accident years, followed by sharp reserve strengthening; in the Allstate case, the 2021 inflation impulse was fully reflected in reserves only in 2022 (strengthening on the order of $1.7 bn). Additionally, it is substantiated that the decisive role is played not by the headline CPI, but by the insurer-specific inflation basket (including repairs, parts, and medical care) and the multiplicative overlay of social inflation, which amplifies pressure in long-tail liability lines. The article may be helpful for actuaries, risk managers, and financial executives of P&C insurers developing robust reserving procedures under non-stationary inflation.

This study investigated the efficiency of separating technological dust generated in cotton ginning plants using two-stage cyclones - SS-6 and SP-1.5. According to practical measurements, it was found that the dust separation efficiency of the SS-6 cyclone was 70.8%, and that of the SP-1.5 cyclone was around 51.3%. The overall efficiency of this two-stage system was 92.5%. At the same time, taking into account high environmental requirements, it is recommended to introduce an additional filtration stage. The three-stage cleaning system allows for more than 99% dust retention, which is important for increasing production efficiency and reducing environmental impact.

Machine learning has emerged as one of the most influential scientific and technological paradigms of the contemporary digital era, reshaping the ways in which data are interpreted, decisions are made, and complex systems are optimized. Across fields ranging from healthcare and robotics to cloud computing and pattern recognition, machine learning techniques have evolved from theoretical constructs into indispensable practical tools. This article presents a comprehensive, theoretically grounded, and application-oriented examination of machine learning, based strictly on established academic and professional references. Drawing from foundational works in probability theory, neural networks, Bayesian learning, and classification theory, as well as more recent contributions related to hybrid systems and applied signal processing, this study develops an integrated view of how different machine learning methodologies complement, compete with, and enhance one another.

The article explores supervised, unsupervised, and hybrid learning paradigms, placing special emphasis on the philosophical and mathematical underpinnings of probabilistic reasoning, evidence weighting, and pattern recognition. It also analyzes decision trees, support vector machines, fuzzy nearest neighbor algorithms, artificial neural networks, and Bayesian classifiers as interconnected components of a broader learning ecosystem rather than isolated techniques. Particular attention is given to the role of hybrid approaches in complex real-world problems, such as electrocardiogram signal analysis and cloud load balancing, demonstrating how machine learning systems can be adapted to deal with uncertainty, noise, and dynamic environments.

Beyond algorithmic descriptions, this work delves deeply into the epistemological foundations of learning from data, examining how evidence, probability, and induction form the basis of predictive modeling. By synthesizing insights from classical and contemporary literature, this article reveals that machine learning is not merely a collection of computational tricks but a coherent scientific discipline grounded in theories of inference, cognition, and decision-making. The discussion also addresses the limitations of existing methods, including issues of overfitting, interpretability, and data dependency, while proposing future directions in hybridization, meta-learning, and adaptive systems. The result is a publication-ready, holistic, and rigorously argued account of machine learning as both a theoretical science and a transformative applied technology.

This article analyzes the physical, mechanical and chemical properties of local coarse wool fiber from a theoretical and methodological perspective. The structure of wool fibers, namely the sheath, core and core layers, their influence on strength and elasticity are scientifically explained. In addition to the thickness, length, twist and strength of wool, its hygroscopicity, thermal conductivity and electrical properties were studied in the study. The technological properties of wool fibers - spinnability, wettability and dyeability - were also extensively analyzed. The fact that wool is grown in different regions, animal breeds, age and climatic conditions directly affect its quality and durability is substantiated with scientific examples. The author shows the moisture properties of wool, its hygroscopicity and comparisons with other textile fibers using clear tables and data. At the same time, modern methods of measuring the strength and tensile strength of wool are also discussed. The results of the research are of practical importance in the processing, cleaning of wool fibers and improving the quality of products made from them. This article serves as a scientific and theoretical basis for the efficient processing of coarse wool resources and the textile industry.

The rapid development of digital technologies has fundamentally transformed the structure and dynamics of the global economy. The digital economy, driven by innovations such as artificial intelligence, big data, cloud computing, and digital platforms, has become a key engine of global economic growth. This paper examines the impact of the digital economy on global economic growth by analyzing its role in increasing productivity, enhancing efficiency, and fostering innovation across various sectors. Digitalization reduces transaction costs, improves access to international markets, and enables new business models that support inclusive and sustainable growth. At the same time, the expansion of the digital economy creates challenges related to digital inequality, labor market transformation, cybersecurity, and regulatory frameworks. The study highlights both the positive contributions and potential risks of digital transformation, emphasizing the need for coordinated policies to maximize economic benefits while minimizing social and economic disparities. The findings suggest that countries investing in digital infrastructure, digital skills, and supportive institutional frameworks are better positioned to achieve long-term economic growth and competitiveness in the global economy.