Бидний тухай
Багш ажилтан
Traditional monetary unit sampling (MUS) in financial auditing leaves detection risk close to 38 per cent while examining only one fifth of the account population, a structural limitation that magnifies the agency problem between public-sector entities and citizens. This study formalizes detection risk as DR = 1 - Recall, bridging machine-learning classifier performance with the ISA 200 audit assurance standard, and proposes a four-layer agentic artificial-intelligence (AI) framework that orchestrates the entire audit workflow without human intervention between successive layers. The framework was applied to 9,907 account-year observations and 3,329,189 individual journal entries (total volume MNT 16.34 trillion) drawn from the general ledger of a large Mongolian state-owned thermal power utility for the fiscal years 2023 to 2025 and was further generalized to seven additional public-sector entities (n = 7,021 accounts). At the account level the ensemble Random-Forest agent reached F1 = 0.978 and reduced detection risk to 2.01 per cent, a 19-fold improvement over MUS. At the transaction level the 15-criterion ensemble flagged 166,459 anomalous entries (5.00 per cent), of which 2,885 constituted very high-risk records. A novel aggregation-masking effect of Benford’s law is reported: 1,247 accounts that appeared statistically normal at the account level concealed clusters of anomalous transactions, a finding only retrievable through dual-level analysis. SHAP-based feature attribution was mapped to a feature-to-ISA ontology (ISA 240, 315, 505, 520) that enabled the autonomous generation of ISA-compliant working-paper text. Six independent validation procedures, including a temporal hold-out, a seven-sector test, and triangulation with two qualified opinions issued by the National Audit Office of Mongolia, confirmed robustness. An open-source web prototype (Audit AI v10.0) demonstrates zero-cost deployability. The study contributes to the broader scientific-culture debate on the ethics, accountability and global human values that should govern the transfer of algorithmic decision-making into public-sector assurance.
The purpose of this study is to analyze the impact of key macroeconomic factors and the time-lag effects of Sustainable Development Goals (SDGs) on their implementation in Mongolia using modern machine learning (ML) techniques. The analysis employs annual data from 2000–2024, including macroeconomic indicators, the SDG Index, and performance metrics for 17 goals. Core variables include GDP growth, inflation, unemployment, exports, fiscal revenue, and government expenditure. In 2025, Mongolia’s SDG Index score reached 66.7, ranking 100th globally, which is close to the regional average but still reflects weak performance in environmental goals. To explain SDG Index fluctuations, four ML models - Linear Regression, Gradient Boosting, Support Vector Regression (SVR), and Artificial Neural Network (ANN) - were compared. Gradient Boosting achieved the highest accuracy (R² = 0.844, RMSE = 0.8735), explaining 84.4% of the variance. SHAP analysis revealed that the most influential factors for SDG performance are health (SDG3), exports, poverty reduction (SDG1), internet usage (SDG9), and food security (SDG2), while inflation and unemployment exert negative effects. Time-lag analysis shows that foundational service goals—SDG3 (health), SDG6 (water and sanitation), SDG11 (urban development), SDG9 (infrastructure), and SDG1 (poverty reduction)—have strong positive one-year lag effects on the SDG Index. In contrast, environmental goals (SDG12–15) exhibit persistent negative lag effects, indicating that ecosystem degradation slows SDG progress in subsequent years. SDG4 (education) and SDG7 (energy) display weak lag impacts, with long-term delayed effects. Additionally, K- Means clustering grouped SDG indicators into three development patterns: (i) economy-driven, (ii) social- service-driven, and (iii) environment-challenged clusters. Findings suggest that improving SDG implementation in Mongolia requires export diversification, investment in health services, inflation control, and strategic reforms in environmental policy.
This study aims to analyze a company’s sales data using data mining techniques to uncover patterns in consumer behavior and product sales trends. Clustering methods (K-Means and K-Nearest Neighbors) and association rules were applied using IBM SPSS Modeler 18.6. The performance and effectiveness of these methods were compared to evaluating their utility in sales data analysis. Clustering results demonstrated the potential to segment consumers based on shared characteristics, thereby supporting data-driven marketing strategy development. Association rule analysis revealed product combinations that are frequently purchased together, offering additional insights to enhance sales strategies. The findings of the study highlight the value of data mining techniques in improving business decision-making and underscore the potential for incorporating larger datasets and more advanced machine learning methods in future analyses.
This research addresses the question of how artificial intelligence (AI) is being adopted in Mongolia’s auditing sector by integrating theoretical approaches with data-driven evidence. Through the application of Partial Least Squares Structural Equation Modeling (PLS-SEM), the research evaluates how auditors’ digital skills, professional qualifications, and personal attributes affect their intention to employ AI in practice. The results highlight digital competence as the most decisive factor, whereas individual traits and expertise have little direct effect. The SWOT analysis points to both opportunities and barriers: while auditors are eager to adopt AI, the lack of training programs, underdeveloped infrastructure, and unclear legal environment slow down progress. This research makes both conceptual and practical contributions by suggesting ways for policymakers, industry leaders, and educational institutions to facilitate AI-enabled auditing in Mongolia.
Тема «Цель устойчивого развития 12: Ответственное потребление и производство — проблемы на примере Монголии» рассматривает влияние современных моделей производства и потребления в Монголии на окружающую среду, экономику и общество. В данной теме особое внимание уделяется таким проблемам, как чрезмерная зависимость экономики от горнодобывающей отрасли, слабая система управления отходами, загрязнение воздуха и почвы, а также неэффективное использование природных ресурсов. Также объясняется необходимость внедрения политики устойчивого производства, переработки отходов и поддержки экологически ответственного потребления. Делается вывод о том, что совместное участие граждан, организаций и государства в рациональном использовании природных ресурсов и развитии экологически чистого производства и потребления имеет важное значение для устойчивого развития Монголии.
The purpose of this study is to analyze the impact of key macroeconomic factors and the time-lag effects of Sustainable Development Goals (SDGs) on their implementation in Mongolia using modern machine learning (ML) techniques. The analysis employs annual data from 2000–2024, including macroeconomic indicators, the SDG Index, and performance metrics for 17 goals. Core variables include GDP growth, inflation, unemployment, exports, fiscal revenue, and government expenditure. In 2025, Mongolia’s SDG Index score reached 66.7, ranking 100th globally, which is close to the regional average but still reflects weak performance in environmental goals. To explain SDG Index fluctuations, four ML models - Linear Regression, Gradient Boosting, Support Vector Regression (SVR), and Artificial Neural Network (ANN) - were compared. Gradient Boosting achieved the highest accuracy (R² = 0.844, RMSE = 0.8735), explaining 84.4% of the variance. SHAP analysis revealed that the most influential factors for SDG performance are health (SDG3), exports, poverty reduction (SDG1), internet usage (SDG9), and food security (SDG2), while inflation and unemployment exert negative effects. Time-lag analysis shows that foundational service goals—SDG3 (health), SDG6 (water and sanitation), SDG11 (urban development), SDG9 (infrastructure), and SDG1 (poverty reduction)—have strong positive one-year lag effects on the SDG Index. In contrast, environmental goals (SDG12–15) exhibit persistent negative lag effects, indicating that ecosystem degradation slows SDG progress in subsequent years. SDG4 (education) and SDG7 (energy) display weak lag impacts, with long-term delayed effects. Additionally, K-Means clustering grouped SDG indicators into three development patterns: (i) economy-driven, (ii) social-service-driven, and (iii) environment-challenged clusters. Findings suggest that improving SDG implementation in Mongolia requires export diversification, investment in health services, inflation control, and strategic reforms in environmental policy.
The purpose of this study is to identify the factors influencing tax revenue collection under the self-assessment system of personal income tax in Mongolia. The study employs the BISEP model, which encompasses Business, Individual, Social, Economic, and psychological factors. Data were collected through a survey of 290 taxpayers in the Sukhbaatar District of Ulaanbaatar, and correlation and regression analyses were conducted. The results indicate that the most significant factors affecting tax revenue collection are the taxpayer’s financial condition and knowledge about taxation (R = 0.66, β = 0.46). In contrast, social and economic factors show relatively weak effects. The psychological aspects and perceptions of fairness among taxpayers were found to have a positive influence on taxpaying behavior. This research provides both theoretical and practical foundations for policymakers to strengthen the voluntary tax compliance system and enhance the effectiveness of tax policy implementation in Mongolia.
In recent years, artificial intelligence (AI) and machine learning (ML) methodologies have rapidly penetrated the fields of auditing, financial analysis, and credit risk assessment, enabling more accurate and real-time evaluations compared to traditional statistical approaches. However, in developing countries such as Mongolia, the integration of these methods into audit and credit evaluation systems remains limited and underexplored. This study aims to develop an integrated model for assessing audit and credit risk and identifying the key influencing factors using machine learning techniques. The analysis is based on data from 88 enterprises that received loans from the Mongolian Small and Medium Enterprise Development Project during 2019–2024, including their financial statements, on- site audit reports, and loan repayment records from the SME Development Fund. Classification algorithms such as Random Forest, Gradient Boosting, and Decision Tree were applied, and their performance was compared using evaluation metrics including Accuracy, Precision, Recall, and F1-score. The results revealed that the Random Forest algorithm achieved the highest performance (Accuracy = 0.944, Recall = 1.000), demonstrating its ability to identify high-risk entities with 100% recall. SHAP analysis indicated that tax arrears, overdue loan days, and non-compliance periods were the most influential variables affecting audit and credit risk. These findings highlight the potential of adopting AI-based integrated risk assessment systems in Mongolia’s auditing and credit supervision sectors, contributing to early risk detection, optimized allocation of supervisory resources, and enhanced transparency at the policy level.
This study examines the effectiveness of sales planning by assessing the proportion of total producible goods that can be successful sold, given constraints related to warehouse capacity, financial resources, and human capital. Using company “MGL AQUA” JSC as a case study, we apply numerical data to develop a goal programming model designed to minimize total deviations from the target function. The model incorporates constraints to ensure that the total cost of sold products does not exceed 13 billion MNT while maintaining a minimum total profit of 16 3 billion MNT. The optimal solution suggests that the planned sales volume should be 30 1 million units. Compared to 2024 this projection represents a 22 5 percent increase in sales performance, with 76 2 percent of the total producible quantity being fulfilled.
The advancement of artificial intelligence (AI) is ushering in a new era in auditing by enhancing decision-making, fraud detection, and risk assessment processes. This study presents a systematic literature review of 30 prominent academic works published between 2017 and 2024. Findings suggest that while AI can significantly improve audit efficiency and accuracy, several challenges remain—such as data security, professional ethics, human factors, and regulatory compliance. The study concludes that a collaborative model, where AI augments rather than replaces auditors, offers a more feasible and ethical approach to AI integration in auditing practices.
In the economic and social policy document to be followed by the government We will develop social services to ensure life security and strengthen the social insurance system to improve the quality of life. Expand the scope of social insurance. Improve service availability. Expanding the middle class of citizens and households by strengthening management. Cultivating proper management of social insurance funds. Bring the fund to a breakeven level. A fully independent social insurance system will be created.
