ConcienciaDigital

Importance of intellectual capital in companies in the agricultural sector

Maribel del Rocío Paredes Cabezas — 2026-07-06

Introduction. Globalization, the internationalization of markets and competition have posed challenges in business management, which must be addressed from different perspectives to be able to successfully overcome them. In this context, intellectual capital emerges as a new concept to be considered, which encompasses knowledge, skills, and attitudes of human talent, essential for decision-making. Objective. The present research aims to determine the importance of intellectual capital in companies in the agricultural sector in the Ambato canton, Tungurahua province. Methodology. The methodology applied consisted of a bibliographic review and the application of the VAIC model to measure intellectual capital in the study sector. Results. The results have allowed us to establish a very strong and positive correlation between ROE and ROA, that is, the Return on Equity (ROE) and the Return on Assets (ROA) maintain a direct relationship, which is consistent given that both indicators reflect the profitability of the company. Conclusion. In conclusion, efficient management of intellectual capital is crucial for the financial success and sustainability of companies. Companies must focus on holistic strategies that integrate the development of human, structural and intellectual capital to improve their performance and competitiveness in the market. General study area: Accounting and Auditing. Specific study area: Intangible assets. Type of study: Original article.

Bayesian nonparametric models in market segmentation: a systematic review and research agenda

Tannia Carolina Calle Monroy — 2026-07-17

Introduction: Bayesian nonparametric (BNP) models have emerged as flexible approaches for clustering and segmentation by allowing data-driven identification of latent structures. Despite their methodological advances, their application in marketing remains limited. Objective: To systematically examine the literature on Bayesian nonparametric (BNP) models for clustering and segmentation, with a focus on their applicability in marketing contexts. Methodology: A PRISMA-based protocol was employed to identify and analyze 43 Scopus-indexed peer-reviewed articles. Bibliometric and thematic analyses were conducted to characterize research trends and methodological developments. Results: The results indicate a marked growth in scientific output, concentrated in statistical and methodological journals, with a strong emphasis on Dirichlet processes, mixture models, and clustering techniques. A taxonomy is proposed that structures the literature into six categories, ranging from foundational BNP models to advanced hierarchical and data-adaptive extensions. Conclusions: The findings reveal a persistent gap between methodological advances and their application in marketing, where traditional clustering approaches prevail. BNP models remain underutilized despite their capacity to capture latent heterogeneity and emergent segmentation structures. General study area: Marketing. Specific study area: Market segmentation, Bayesian nonparametric models, clustering. Article type: Systematic literature review.

Predictive models of meteorological data through machine learning and their impact on the agroproduction of the Bolívar province, Ecuador

Deysi Margoth Guanga Chunata — 2026-07-17

Introduction. The inherent non-stationarity of Andean climate series limits the efficacy of classical linear models. Objective. To develop a comparative methodological framework integrating traditional statistical models (ARIMA/SARIMA) with Ensemble Learning algorithms (Random Forest, XGBoost) and deep neural networks (LSTM) for meteorological variable prediction, evaluating their causal impact on agroproduction via VAR modeling. Methodology. Quantitative longitudinal study with 1,642,217 observations (2016-2023). Robust preprocessing (Isolation Forest, KNN Imputer) and walk-forward temporal validation were applied. Multivariate analysis utilized unit root tests (ADF), Granger causality, and Forecast Error Variance Decomposition (FEVD). Results. Random Forest significantly outperformed SARIMA (RMSE 0.74 vs 1.66; R² 0.67 vs -0.63). The VAR (6) model evidenced significant Granger causality (p < 0, 001) between precipitation/temperature and corn yield, with a dynamic lag of 6 months. Conclusion. Non-parametric methods better capture the conditional heteroscedasticity of climate data, providing robust tools for precision agriculture. General Area of Study: Data Science and Artificial Intelligence. Specific area of study: Agroclimatic predictive modeling. Type of study: Original articles.