NMNataliia K. Maksyshko

Nataliia K. Maksyshko

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Nataliia K. Maksyshko has a total of 44 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Modelling the efficiency of the agricultural sectors in EU countries and Ukraine: integrated assessment, convergence analysis, and typology

    by Bohdan Kyshakevych, Olga Melnyk, Oksana Yurynets, Nataliia Maksyshko, Yaroslav Kotyk

    Purpose. The study aims to develop an integrated approach to assessing the efficiency of the agricultural sectors of EU countries and Ukraine and to examine their convergence patterns, stability, and development trajectories. It also seeks to determine whether lower-efficiency countries demonstrate catching-up dynamics, whether cross-country disparities diminish over time, and how Ukraine is positioned within the European agricultural efficiency framework. Methodology. The integrated efficiency index combines normalised CRS-DEA, VRS-DEA, SFA, and Malmquist total factor productivity change estimates. CRITIC is used as the principal objective weighting method, while entropy weighting provides a robustness check. β- and σ-convergence models, correlation analysis, and criterion-based typological classification are applied to EU data for 2016–2024; Ukraine is assessed for 2016–2022 because comparable later data are unavailable. Results. The findings reveal substantial cross-country differences in agricultural efficiency, with Denmark, the Netherlands, Belgium, and Italy among the leading countries, while Ukraine remains in the low or unstable efficiency group. The results confirm predominantly conditional β-convergence and partial, non-monotonic σ-convergence. Higher average efficiency is strongly associated with lower volatility, while CRITIC- and entropy-based country rankings are highly consistent. Four distinct efficiency trajectory groups are identified. Originality. The study integrates deterministic and stochastic frontier methods, productivity dynamics, objective weighting, convergence analysis, and country typology within a unified analytical framework. Unlike approaches based on a single efficiency measure or predefined weights, the proposed methodology combines static and dynamic dimensions of agricultural performance using data-driven weighting. The integrated index is applied not only to country ranking but also to analysing convergence, stability, and heterogeneous development trajectories. Practical implications. The results provide an analytical basis for differentiated agricultural policies tailored to the efficiency level, dynamics, and stability of each country group. Policy priorities include maintaining technological leadership in highly efficient countries, supporting modernisation in catching-up economies, and strengthening structural and institutional support in low or unstable efficiency systems.

  • Open Access

    APPLICATION OF ARTIFICIAL INTELLIGENCE TOOLS FOR FORMING THE SAFETY STOCK MATRIX IN RETAIL TRADE: PROMPT ENGINEERING METHODOLOGY

    by S. S. Cheverda, N. K. Maksyshko, S. M. Butenko, I. О. Dubyna

    The article investigates the practical foundations of applying large language models (LLM) of artificial intelligence for automated formation of a safety stock matrix in small retail trade. The object of the study is the «Kamilla» retail network (Tiachiv, Zakarpattia region) – five household chemicals and hygiene goods stores with a centralized warehouse infrastructure. It is established that approximately 80% of the network’s assortment positions are characterized by intermittent (irregular) demand, which makes effective application of classical time series forecasting methods impossible and necessitates adaptive analytical tools. It is shown that the transition from decentralized to centralized product placement increases the inventory utilization ratio from 53% to 71%, releasing over UAH 1 million of frozen working capital. A comparative testing of four free LLM platforms – Microsoft Copilot, OpenAI ChatGPT, Google Gemini, and Anthropic Claude – was conducted using a unified query. It is established that the Claude Sonnet 4.6 model in Extended Thinking mode demonstrates the highest analytical efficiency for structured business data. A three-stage methodology of iterative prompt engineering has been developed and tested: structuring instructions, delegating the choice of forecasting method to the model depending on the time series type, and implementing cross-verification mechanisms for result completeness. A cloud solution architecture based on the integration of Claude AI and Google Drive is proposed, which reduces the weekly order formation cycle to 3–5 minutes without direct financial implementation costs. The practical value of the study lies in providing small business managers with a systematic, reproducible, and zero-cost framework for inventory decision support. Стаття присвячена розробці та апробації методології застосування великих мовних моделей (LLM) штучного інтелекту для автоматизованого формування матриці страхових запасів у малій роздрібній торгівлі. На прикладі торговельної мережі «Камілла» встановлено, що близько 80% асортиментних позицій характеризуються інтермітуючим (нерегулярним) попитом, що унеможливлює ефективне застосування класичних методів прогнозування. Показано, що перехід від децентралізованого до централізованого розміщення товарів підвищує коефіцієнт реалізації запасів із 53% до 71%, вивільняючи понад 1 млн грн. заморожених оборотних коштів. Здійснено порівняльне тестування чотирьох безкоштовних LLM-платформ: Copilot, ChatGPT, Gemini та Claude. Встановлено, що модель Claude Sonnet 4.6 у режимі Extended Thinking демонструє найвищу аналітичну ефективність для структурованих бізнес-даних. Розроблено трьохетапну методологію ітеративного prompt engineering: структурування інструкцій, делегування вибору методу прогнозування моделі та впровадження механізмів крос-верифікації. Запропоновано хмарну архітектуру рішення на базі інтеграції Claude AI та Google Drive, яка скорочує щотижневий цикл формування замовлень до 3–5 хвилин без прямих фінансових витрат.

  • Open Access

    Modelling the impact of state support types on agricultural production, considering exogenous shock and cyclical factors

    by Roman Ivanov, Yurii Hurtovyi, Nataliia Maksyshko, Volodymyr Katan, Mykola Osypchuk

    Purpose. The purpose of the study is to empirically assess the impact of individual measures of state support for agriculture in the EU, Ukraine, and New Zealand, considering production and market factors and external challenges that determine the value of agricultural products. Methodology / approach. To achieve the goal, production function models were used in six successive modifications, considering (1) the rate of technological progress, (2) total budget support for agriculture, (3) individual measures of state support determined as statistically significant, and (4) their combination with land resources – agricultural area, (5) removed statistically insignificant parameters, (6) specific adjustment variables. To verify the reliability and accuracy of the models built based on data for 2010–2022, they are tested by forecasting the value of agricultural products for 2023, and the autocorrelation of the residuals is estimated using the Ljung-Box test. Results. The study proves that the effectiveness and impact of individual state support measures on the formation of agricultural product value differs significantly between countries. While the financing of agricultural knowledge and innovation is one of the key development drivers in the EU, inspection and controls are a growth stimulator in New Zealand. At the same time, they act as a barrier in Ukraine, demonstrating negative elasticity. The inclusion of adjustment variables allows for a quantitative estimation of the negative impact of exogenous shocks, the COVID-19 pandemic, and the Russian-Ukrainian war. The analysis of production factors demonstrates the pivotal role of capital investments, effective use of irrigated land, yield, as well as foreign trade factors for Ukraine and New Zealand, which have export-oriented agricultural production. Originality / scientific novelty. For the first time, the proposed models integrate explanatory variables that reflect large-scale exogenous shocks, namely the macroeconomic impact of COVID-19, considering demographic losses and the gradual fading of the pandemic in the EU, an integrated indicator of the destructive consequences of the war, which includes demographic casualties, destruction of irrigation and energy infrastructure in Ukraine. Practical value / implications. The results obtained have practical significance for improving the mechanisms of agricultural policy aimed at increasing production results and the sustainability of agriculture in the face of global and local crises.


  • Open Access

    Systemic issues and efficiency reserves in EU agriculture: a slack-based DEA approach

    by Bohdan Kyshakevych, Olga Melnyk, Natalia Maksyshko, Yuriy Maturin, Yaroslaw Kotyk

    Purpose. The purpose of the study is to develop an integrated approach for evaluating the super-efficiency of the agricultural sector in EU countries, considering resource, ecological, and economic dimensions, while identifying systemic challenges and opportunities for efficiency enhancement through slack-based Data Envelopment Analysis (DEA). Methodology / approach. The study uses super efficiency slack-based Data Envelopment Analysis models to evaluate the efficiency of the agricultural sector in EU countries. It incorporates three input-oriented DEA models focusing on resource-based, ecological, and economic dimensions to provide a comprehensive assessment. An aggregated assessment of the efficiency of the agricultural sector in EU countries was proposed based on the geometric mean of the evaluations from the constructed DEA models. A method for identifying systemic issues and opportunities for efficiency improvements in the agricultural sector of 27 EU countries using slack values was proposed. Results. The most efficient countries in the EU agricultural sector according to proposed approach on the base of aggregation of three offered DEA model are the Netherlands, Belgium, Cyprus, Malta, Denmark, France and Ireland, demonstrating high levels of super-efficiency. The main challenges for less efficient countries include excessive land use, low export levels, and insufficient added value of agricultural products. Slack analysis revealed that the largest deviations are observed in land resource utilisation and export volumes. Originality / scientific novelty. The article proposes an approach for evaluating the super-efficiency of the agricultural sector in EU countries by aggregating the assessments of three proposed DEA models. A method for identifying systemic issues in the efficient use of resources in the agricultural sector based on the analysis of slacks in DEA models has been proposed. Practical value / implications. The article provides practical insights for policymakers and stakeholders to optimise resource use and improve the efficiency of the agricultural sector in EU. The results can guide the development of tailored strategies to address inefficiencies, particularly in land use and export performance. The proposed approach provides a methodological framework that can be applied to other regions, promoting sustainable agricultural practices globally and enabling the development of an aggregated efficiency ranking for the agricultural sector.


  • Open Access

    ELECTRIC VEHICLE SALES FORECASTING: COMPARATIVE ANALYSIS OF CLASSICAL AND GENERATIVE MODELS

    by S.G. Kurkula, N.K. Maksyshko, D.V. Ocheretin, O.A. Raksha

    The article compares the effectiveness of classical statistical methods, machine learning algorithms, and the modern generative model TimeGPT in forecasting electric vehicle sales in 22 countries worldwide. The study covers the following models: ARIMA, ARIMA Boosted, Prophet, Random Forest, linear regression, and TimeGPT, which are tested on a single dataset with subsequent determination of the most effective model for each market. The research results confirmed differences in the behavior of various models in the presence of seasonality and nonlinear dynamics, as well as the impact of the time series length on the model’s behavior. The results obtained allow us to formulate practical recommendations for the selection of forecasting methods, taking into account seasonality, nonlinear trends, data volume, and market specifics. This creates a basis for improving the accuracy of forecasts, optimizing production and investment planning, and supporting the strategic development of the electric vehicle market. The proposed approach can be expanded in future research through the integration of multifactor data models, scenario modeling, and model architecture improvements. Стаття присвячена порівнянню ефективності класичних статистичних методів, алгоритмів машинного навчання та сучасної генеративної моделі TimeGPT у задачі прогнозування обсягів продажів електромобілів на ринках 22 країн світу. Дослідження охоплює такі моделі: ARIMA, ARIMA Boosted, Prophet, Random Forest, лінійна регресія та TimeGPT, які випробовуються на єдиному наборі даних із подальшим визначенням найбільш результативної моделі для кожного ринку. Результати дослідження підтвердили відмінності у поведінці різних моделей за наявності сезонності та нелінійної динаміки, а також як довжина часового ряду впливає на поведінку моделі. Отримані результати дозволяють сформувати практичні рекомендації щодо вибору методів прогнозування з урахуванням сезонності, нелінійності трендів, обсягу даних та ринкової специфіки. Це створює основу для підвищення точності прогнозів, оптимізації виробничого та інвестиційного планування, а також підтримує стратегічний розвиток ринку електромобілів. Запропонований підхід може бути розширений у майбутніх дослідженнях через інтеграцію багатофакторних моделей даних, сценарного моделювання та вдосконалення архітектури моделей.

  • Open Access

    CONCEPTUAL MODEL OF ARTIFICIAL INTELLIGENCE APPLICATION IN VIDEO GAME DEVELOPMENT PROJECT MANAGEMENT

    by S.S. Cheverda, N.K. Maksyshko, O.I. Bashtannyk

    The article focuses on developing a conceptual model for artificial intelligence application in video game development project management. It examines the complex relationships between game product creation stages and corresponding AI tools that optimize production processes. The research substantiates the relevance of this issue through growing competition in the video game market, dynamic technological changes, and continuously evolving user needs. The paper identifies and analyzes in detail three fundamental principles of artificial intelligence integration: automation of routine processes through automated testing and data analysis tools; strategic planning and resource management to forecast project needs and optimize task distribution; and improving user interaction through real-time feedback collection and analysis. The conceptual framework presented by the authors demonstrates the step-by-step integration of specific artificial intelligence tools, including ChatGPT, DALL-E, Unity ML Agents, IBM Watson, Nvidia DLSS, GameAnalytics and others, into development blocks – from idea generation and initial planning to testing, release, and user feedback analysis. Special attention is paid to mechanisms of testing process automation, resource planning optimization, market trend forecasting, and analysis of player behavioral patterns. The paper describes AI application at each stage of the production cycle, noting specific benefits that development teams receive through the implementation of these technologies. Research results confirm that systematic AI implementation contributes significantly to improving development efficiency, end product quality, risk reduction, and resource cost optimization. The developed conceptual model can serve as a theoretical foundation for creating methodological recommendations on AI integration into video game development project management processes, opening new opportunities for enhancing product competitiveness in this dynamic industry. Стаття присвячена розробці концептуальної моделі застосування штучного інтелекту в управлінні проєктами розробки відеоігор. Досліджено комплексні взаємозв’язки між етапами створення ігрових продуктів та відповідними інструментами ШІ, що дозволяють оптимізувати виробничі процеси. У дослідженні обґрунтовано актуальність проблематики зростаючою конкуренцією на ринку відеоігор, динамічними змінами технологій та постійно еволюціонуючими потребами користувачів. У роботі визначено та детально проаналізовано три фундаментальні положення інтеграції штучного інтелекту: автоматизація рутинних процесів через інструменти автоматизованого тестування та аналізу даних; стратегічне планування та управління ресурсами з метою прогнозування потреб проєкту та оптимізації розподілу завдань; покращення взаємодії з користувачами через збирання та аналіз зворотного зв’язку в режимі реального часу.

  • Open Access

    CONCEPTUAL MODEL FOR FORECASTING THE DYNAMICS OF ELECTRIC VEHICLE SALES VOLUMES ON THE GLOBAL MARKET

    by S.G. Kurkula, N.K. Maksyshko

    The modern electric vehicle market is characterized by rapid development and significant fluctuations driven by technological, economic, and political factors. Forecasting sales dynamics is critically important for the strategic planning of manufacturers, investors, and government institutions. However, existing forecasting methods often lack accuracy due to data heterogeneity, insufficient consideration of market-specific features, and influencing factors. This paper proposes a conceptual model for forecasting electric vehicle sales dynamics in the global market. The model incorporates market clustering, time series analysis, and both causal and non-causal forecasting models. The model includes the following stages: data collection, clustering of markets based on dynamic characteristics, construction of causal (regression model using an MLP neural network) and non-causal (ARIMA, RNN, hybrid model) forecasting models, as well as evaluation of forecast quality and selection of the most relevant model. The scientific novelty of the research lies in a comprehensive approach that combines modern forecasting methods, enabling improved prediction outcomes in an emergent economy. The proposed model demonstrates potential due to its use of hybrid methods and the integration of neural networks alongside traditional statistical approaches. The research results may be valuable for manufacturers, investors, and government bodies in planning infrastructure projects, developing support policies, and assessing market trends. The model also opens up opportunities for further research in the field of forecasting dynamic markets. Сучасний ринок електромобілів характеризується стрімким розвитком та значними коливаннями, зумовленими технологічними, економічними та політичними факторами. Прогнозування динаміки продажів є критично важливим для стратегічного планування виробників, інвесторів та державних установ. Однак існуючі методи прогнозування часто мають обмежену точність через неоднорідність даних, недостатнє врахування специфіки окремих ринків та факторів що на них впливають. У цій статті запропоновано концептуальну модель прогнозування динаміки обсягів продажів електромобілів на світовому ринку, яка включає кластеризацію ринків, аналіз часових рядів, каузальні та некаузальні моделі прогнозування. Модель включає такі етапи: збір даних, кластеризацію ринків за характером динаміки, побудову каузальних (регресійна модель за допомогою нейронної мережі MLP) та некаузальних (ARIMA, RNN, гібридна модель) моделей, а також оцінку якості отриманих прогнозів та вибір найбільш релевантної моделі. Наукова новизна дослідження полягає в комплексному підході, який поєдную сучасні методи прогнозування, що дозволить покращити результати прогнозів в умовах емерджентної економіки. Запропонована модель має потенціал завдяки використанню гібридних методів та використанню нейронних мереж наряду із традиційними статистичними підходами. Результати дослідження можуть бути корисними для виробників, інвесторів та державних органів при плануванні інфраструктурних проектів, розробці політик підтримки та оцінці ринкових тенденцій. Модель також відкриває перспективи для подальших досліджень у галузі прогнозування динамічних ринків.

  • Open Access

    ANALYZING THE EFFICIENCY OF DIGITALIZATION IN SMALL AND MEDIUM-SIZED ENTERPRISES ACROSS EU COUNTRIES USING DEA MODELS

    by Bohdan Kyshakevych, Natalia Maksyshko, Kostiantyn Hrytsenko, Ivan Voronchak, Bohdan Demediuk

    The process of digitizing businesses involves ongoing investments and additional costs, leading to the inevitable inquiry into the economic impact of digital transformation. Specifically, it's important to evaluate how this process contributes to the economic growth of nations. The article introduces a method for creating a ranking of European countries according to the economic benefits derived from digitalizing businesses of different scales. To achieve this, DEA models were developed to measure the effectiveness of digital adoption by small and medium-sized enterprises and larger corporations. For input variables, we selected metrics that reflect the extent of business digitalization, specifically, the proportion of SMEs (small and medium-sized enterprises) with a high digital intensity index (DII), the percentage of SMEs utilizing cloud technologies, and the contribution of SME turnover to the overall e-commerce turnover. The output variables chosen were metrics indicative of a country's economic health: GDP per capita, the ratio of exports to GDP, and the international investment position. The evidence demonstrates that due to significant differences in size, economic advancement, and digitalization levels among EU countries, the VRS model more adeptly accommodates these disparities. This approach permits the evaluation of each country's performance based on its unique situation and potential for development, without consideration of its economic magnitude. In 2022, Denmark, the Netherlands, Sweden, Finland, Malta, and Belgium emerged as the frontrunners in digital intensity among SMEs, with nearly 40% of businesses exhibiting a high degree of digital engagement. Among these leaders, only Malta and the Netherlands achieved the highest efficiency in leveraging digitalization for economic growth, as determined by the VRS model that considers variable returns to scale. It was shown, that the economic effect of digital transformation on large enterprises in EU countries is more substantial than on small and medium-sized enterprises.


  • Open Access

    MATHEMATICAL MODEL OF THE PRODUCT SUPPLY PROCESS WITH UNCERTAIN DELIVERY PRICES

    by I.V. Kozin, N.K. Maksyshko, K.O. Donina

    The article addresses the issue of product delivery under the condition that delivery costs are estimated based on the reports of economic agents. The cost of delivering goods along a specific route may be uncertain and dependent on the individual preferences of carriers. For instance, a carrier might prefer a route that passes by a store they want to visit, a dining facility, or one that is perceived as safer from the carrier’s perspective. Naturally, only the carrier can determine this internal cost of transporting goods. The question arises: how can a system of delivery routes be chosen that serves as a compromise between the carriers’ preferences? Formally, the task is to minimize the collective transportation costs as much as possible. Here, the costs are calculated based on the prices reported by individual economic agents, the carriers. However, the carriers’ reports may not necessarily be truthful. To secure a more favorable route, a carrier might provide false information about transportation costs. How can agents be incentivized to provide truthful information? One approach to solving this problem is the key agent mechanism, which is discussed in this paper. As a result of the study, a mathematical model of the product supply process with uncertain delivery prices has been developed, which can be used in the development of decision support systems in transportation logistics. Стаття присвячена проблемі доставки продукції при умові, що вартість доставки оцінюється на базі повідомлень економічних агентів. Вартість доставки продукції за конкретним маршрутом може бути невизначеною і базуватися на індивідуальних уподобаннях перевізників. Наприклад, перевізник може прагнути вибрати маршрут, який проходить повз магазин, який він хотів відвідати або повз їдальні, або просто безпечніший з точки зору конкретного перевізника. Звісно, цю внутрішню ціну перевезення товару з погляду окремого перевізника може визначити лише сам перевізник. Виникає питання: як вибрати таку систему маршрутів доставки в транспортній задачі, яка була б своєрідним компромісом між побажаннями перевізників. Формально завдання полягає в тому, щоб якомога зменшити колективні витрати на перевезення товару. Тут витрати розраховуються виходячи з цін, які повідомляють окремі економічні агенти – перевізники. Проте повідомлення перевізників не обов’язково будуть правдивими. Для того, щоб забезпечити собі більш вигідний маршрут, перевізник може передати хибне повідомлення про ціну перевезення. Як змусити агентів повідомляти правдиву інформацію? Одним із способів вирішення такого завдання є механізм ключових агентів, який і розглядатиметься в даній роботі.


  • Open Access

    Model aspect of the study of the processes of sustainable development of socio-economic systems

    by R V Ivanov, T V Grynko, V M Porokhnya, N K Maksyshko, V V Oglih

    The triune concept of sustainable development takes into account environmental, social and economic issues in general. At the same time, the solution of the problems of resource conservation and socio-economic development do not contradict each other, but contribute to mutual reinforcement. The purpose of this study is to build, describe, test and analyze economic and mathematical models of the processes of self-organization of the socio-economic system within the framework of the concept of sustainable development. The considered models take into account both quantitative and qualitative characteristics. The research methodology is based on the assumption that the system under consideration is dynamic. The study makes it possible to evaluate the process of sustainable development of the socio-economic system in terms of the formation and influence of its characteristics on the overall result. Various options for formalizing the level of motivation of economic activity are described and analyzed, one of which is based on the economic interpretation of the Weber-Fechner psychophysiological law. The assumption of the continuous development of innovative technologies made it possible to reveal the existence of a “fatigue point”, the onset of which indicates the possibility of increasing the growth rate of production activities due to its technological transformation.