Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Ángel Iván Torres Quijije has a total of 34 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Forecastability of Intentional Homicide Patterns in Ecuador (2014–2026): A Rolling-Origin Evaluation of Statistical Forecasting Algorithms

    by Angel Ivan Torres-Quijije, Manuel Sarmiento-Fajardo, Emilio Zhuma-Mera, Byron Oviedo-Bayas

    Ecuador has experienced one of the fastest escalations of lethal violence in Latin America, with annual intentional homicides rising from about 1000 in 2014–2019 to 9283 in 2025. This study evaluates how well such patterns can be forecast, and what governs that forecastability, using the complete official open dataset of 43,976 victim-level records covering January 2014–June 2026 (portal update 15 July 2026). Incidents are aggregated into four monthly pattern series—case volume, firearm share, mean victim age and geographic concentration in Guayas—and fifteen specifications—classical, automatically configured, and state-of-the-art neural and gradient-boosted baselines—are compared under a rolling-origin protocol with 31 evaluation origins and a 12-month horizon, with every configuration decision taken inside each training window. Accuracy is reported as the mean and standard deviation of MAE, RMSE, MAPE, MASE and R2 across origins, interval quality as empirical coverage and the mean scaled interval score, and differences are tested with Diebold–Mariano statistics under Holm correction. No pairwise comparison among the 303 tested survives multiplicity control, and in all four series, the best and second-best specifications are statistically indistinguishable: algorithm choice is not what these data can resolve, and the result holds with current deep-learning architectures in the candidate set rather than in their absence. The test is conservative and its power is limited, so this is non-detectability at this sample size rather than evidence that the differences are zero. What does determine it are diagnosable properties of the series. Bounded, low-variability series are forecast with mean MASE between 0.58 and 1.13 across the full candidate set, and below 1.00 by twelve of the fifteen specifications on the firearm share, whereas unbounded count series under regime change exceed an MASE of 2.09 for every specification tested, including simple benchmarks. No series requires a seasonal difference—the OCSB test returns D=0 in all four and the automatic search never selects one—although weak seasonal autoregressive terms are retained at some origins, and the firearm share shows a small but significant monthly effect. A genuine external validation against the observed 18 months of 2025–2026, unseen by any model, reproduces the same asymmetry (firearm share, MAPE 2.3% with 100% interval coverage; volume, bias −14.0%). Splitting the evaluation by regime shows that the crisis made count series roughly twice as unpredictable and bounded series roughly twice as predictable: the escalation destabilised the level of the phenomenon while stabilising its composition.

  • Open Access

    Comparative Evaluation of SARIMA, Prophet, and Holt–Winters Algorithms for Forecasting Intentional Homicide Patterns in Ecuador (2020–2024)

    by Angel Torres-Quijije, Manuel Sarmiento-Fajardo, Emilio Zhuma-Mera, Byron Oviedo-Bayas

    Ecuador experienced one of the fastest escalations of lethal violence in Latin America, with the intentional homicide rate rising from 7.8 per 100,000 inhabitants in 2020 to 47.25 in 2023. This study develops a replicable, algorithm-selection framework for forecasting the temporal evolution of four key patterns of intentional homicide (monthly case volume, firearm share, mean victim age, and geographic concentration in the province of Guayas) using official open data comprising 24,000+ incident records (January 2020–December 2024). Individual incidents were aggregated into monthly time series, and three forecasting algorithms with distinct inductive biases (SARIMA, Prophet, and Holt–Winters exponential smoothing) were compared under an out-of-sample protocol that reserves the 12 months of 2024 as a test set, using MAE, RMSE, and MAPE. No single algorithm dominated: SARIMA was the most accurate for the volume and geographic series, Holt–Winters for the firearm-share series (MAPE 2.85%), and Prophet for the volatile mean-age series (MAPE 5.73%). Re-trained on the full sample, the selected models project a five-year horizon (2025–2029) in which homicides stabilize at historically high levels, roughly four times the 2020 baseline, with firearm involvement consolidating above 85% and violence remaining concentrated in Guayas. The results support pattern-specific algorithm selection for criminological forecasting and provide a quantitative baseline for security policy evaluation in Ecuador.

  • Open Access

    Propuesta para el aprendizaje de división de redes en subredes utilizando gamificación

    by Emilio Zhuma Mera, Orlando Erazo Mera, Ángel Torres Quijije, Brithany Herrera Romero

    Este estudio aborda la necesidad de métodos pedagógicos innovadores en la enseñanza de redes, mediante la integración de la gamificación para la división de redes en subredes y el uso de máscaras de subred. Se implementaron estrategias lúdicas dentro del currículo teórico, demostrando una mejora notable en la comprensión y retención de los estudiantes, como se evidenció en los resultados de las evaluaciones. La participación y el entusiasmo estudiantil incrementaron, destacando la efectividad del enfoque interactivo. Este aporte pedagógico marca un significativo avance, alineándose con las tendencias educativas actuales y promoviendo un aprendizaje más dinámico y efectivo.

  • Open Access

    Desempeño de redes SD-WAN vs. MPLS

    by Santiago Meneses Narvaez, Emilio Zhuma Mera, Ángel Torres Quijije, Marco Chimbo Fogacho

    Software-Defined Networks (SDN) have emerged to enhance flexibility and agility in network environments by enabling management through the separation of the control plane and the forwarding plane. This separation allows networks to quickly adapt to changing environments. Meanwhile, Multiprotocol Label Switching (MPLS) networks manage traffic using a label-switching mechanism, providing key advantages in enterprise environments. This research comparatively analyzes the performance of Software-Defined Wide Area Networks (SD-WAN) and Multiprotocol Label Switching (MPLS) networks, focusing on critical parameters such as latency, jitter, bandwidth, and packet loss. For the analysis, different simulation scenarios have been designed, each subjected to a controlled traffic load. The results of each generated load are sent to a database, which is then connected to Grafana, allowing for easy data visualization through graphs.

  • Open Access

    Sistema telemático con asistente conversacional basado en un modelo de lenguaje para la gestión de citas psicológicas en una universidad ecuatoriana

    by Paola Benítez Navarrete, Ángel Torres-Quijije, Stalin Carreño Sandoya

    Introducción: la gestión de citas del servicio de Psicología de la Universidad Técnica Estatal de Quevedo (Ecuador) dependía de canales presenciales y de mensajería. El objetivo fue diseñar, implementar y evaluar un sistema telemático con un asistente conversacional basado en un modelo de lenguaje. Metodología: estudio aplicado y cuantitativo, con una fase diagnóstica transversal (n = 30) y una evaluación descriptiva posterior a la implementación (n = 20), en grupos independientes seleccionados por conveniencia y sin grupo de control. El sistema integra una arquitectura web de tres capas (PHP, MySQL), un calendario en tiempo real y un asistente GPT-4o-mini con llamada a funciones. Se aplicaron pruebas de tareas y cuestionarios Likert; se calcularon proporciones con intervalos de Wilson y pruebas de Wilcoxon. Resultados: en el diagnóstico, el 86,2 % solicitaba citas presencialmente y el 58,6 % señaló fechas lejanas como principal dificultad. En la evaluación, el 95 % completó la reserva sin asistencia (IC 95 %: 76,4–99,1), todos completaron la reprogramación y la cancelación, el tiempo medio de agendamiento fue de 2,4 minutos, el asistente registró un 92 % de respuestas correctas y la satisfacción media fue 4,85/5 (W+ = 190; p < 0,001). Conclusiones: los resultados describen un funcionamiento favorable y una buena aceptación del sistema, sin permitir inferencias causales. Su uso responsable requiere restringir el asistente a tareas administrativas, minimizar los datos y mantener la supervisión humana, conforme a la Ley Orgánica de Protección de Datos Personales.

  • Open Access

    Diagnóstico de la enseñanza de las matemáticas en octavo año y propuesta de aprendizaje basado en problemas en una institución rural ecuatoriana

    by Jimmy Aldrin Cedeño Barzola, Adriana Carolina Vélez Sánchez, Ángel Iván Torres Quijije

    Introducción: El bajo rendimiento en matemáticas es un problema persistente en la educación básica ecuatoriana, más aún en zonas rurales con pocos recursos. Este estudio caracterizó la enseñanza de la asignatura en octavo año de una institución rural de El Empalme (Ecuador) y, a partir de ese diagnóstico, diseñó una propuesta de aprendizaje basado en problemas (ABP). Metodología: Estudio de enfoque mixto, alcance descriptivo y fase propositiva. Participó toda la población vinculada al área: 43 estudiantes, tres docentes de Matemáticas y el rector. Se aplicaron encuestas a estudiantes y docentes, entrevistas semiestructuradas, observación de aula e inventario de recursos. Los datos se analizaron con frecuencias y categorización temática, y se triangularon entre fuentes. Resultados: El 72,1 % de los estudiantes afirmó que nunca comprende con facilidad las clases y el 69,8 % que nunca se siente motivado. El 90,7 % declaró no entender el concepto de número entero y el 88,4 % el de fracción. Dos de los tres docentes señalaron que la enseñanza privilegia la memorización. La institución dispone de recursos mínimos y el rector avaló la adopción del ABP. Con esta base se diseñó una planificación de seis sesiones con problemas del entorno agrícola y comercial. Conclusiones: El diagnóstico justifica una transición hacia metodologías activas. La propuesta es viable con los recursos disponibles; su efecto sobre el rendimiento deberá evaluarse en una fase de implementación.

  • Open Access

    Eficiencia energética en sistemas IoT sobre redes 5G: estudio comparativo por simulación de modos de ahorro de energía en tres escenarios de aplicación

    by Juan Carlos Pisco Vanegas, Diego Fernando Intriago Rodríguez, Julius Manuel Vinueza Loor, Ángel Iván Torres Quijije

    Introducción: El despliegue masivo de dispositivos del Internet de las cosas (IoT) sobre redes 5G plantea retos energéticos, y faltan comparaciones reproducibles de los modos de ahorro entre dominios. Metodología: Se simularon en MATLAB/GNU Octave tres escenarios (hogar inteligente, salud y agricultura), cada uno con 1 024 dispositivos RedCap distribuidos en ocho celdas NR de 3.5 GHz; cada celda es servida por un gNodeB con un arreglo de 128 antenas. La potencia del módem se modeló con 3GPP TR 38.840. El tráfico de hogar y salud se parametrizó con conjuntos de datos públicos, y el de agricultura con periodos supuestos. Se compararon una línea base y cuatro modos de ahorro con SNR de 20 dB y 40 dB, en 10 corridas (media ± DE). Resultados: La suspensión logró la mayor reducción de energía (92.81 ± 0.20 %, 89.58 ± 0.26 % y 93.40 ± 0.35 % con SNR = 20 dB), con latencia media de 67 ms. La suspensión profunda se limitó a 59–76 % por el costo del establecimiento de conexión. En los perfiles evaluados, convino la suspensión profunda con periodos de 170 s o más, la suspensión entre 1.4 s y 60 s y la espera con periodos de 1 s o menos. Con el canal AWGN simplificado, la SNR alteró el consumo en menos de 0.1 %. Conclusiones: En las condiciones del modelo, el modo óptimo dependió del periodo de transmisión y del presupuesto de latencia de cada clase, más que de la calidad del canal. Palabras clave: Gestión energética; MIMO masivo; redes 5G; Internet de las cosas; simulación de redes.

  • Frequency Allocation Techniques in Fifth-Generation Mobile Networks Based on Artificial Intelligence

    by Diego Fernando Intriago Rodríguez, Diego Fernando Intriago, Wimper Josue Triana Cuenca, Angel Ivan Torres Quijije, Paola Maribel Benítez Navarrete

  • Open Access

    CLINICAL VALIDATION OF A NON-INVASIVE SYSTEM FOR MONITORING PHYSIOLOGICAL PARAMETERS IN HUMANS BASED ON THE INTERNET OF THINGS AND MACHINE LEARNING

    by Emilio Rodrigo Zhuma Mera, Ángel Torres Quijie, Geovanny José Brito Casanova, Jefferson Alexander Moposita Saquinga

    The monitoring of physiological parameters in individuals undergoing frequent clinical follow-up continues to rely, in many cases, on invasive procedures or equipment that makes continuous monitoring outside the hospital setting difficult. In light of this situation, the present study evaluated a non-invasive system for measuring glucose, oxygen saturation, heart rate, and body temperature, supported by the Internet of Things and machine learning. The research was conducted in 2025 at a private clinic in the Buena Fe canton, Los Ríos province, Ecuador, using a quantitative, exploratory, descriptive, and quasi-experimental design, with a sample of 120 patients selected from a population of 174 users. Data collection was performed under controlled conditions, and each measurement obtained by the system was compared with values from commercial reference devices. For predictive processing, models such as linear regression, support vector machines, decision trees, and neural networks were analyzed and evaluated using performance metrics including R², MSE, RMSE, accuracy, F1-score, and AUC. The findings showed that the system performed well in estimating physiological signals, with better performance in the more complex machine learning-based models. The neural network achieved the highest performance indicators, with an accuracy of 95.42% and an F1-score of 98.13%, while the system exhibited low percentage errors in the clinical comparison for oxygen saturation. It is concluded that non- invasive physiological monitoring supported by IoT and machine learning constitutes a viable option for continuous clinical monitoring, with potential for use in telemedicine and in healthcare settings with limited resources.

  • Identification of Patterns for Technological Incident Management in a Telecommunications Company, Period 2021–2024

    by Angel Ivan Torres Quijije, Anthony Jahir López Coello, Diego Fernando Intriago Rodriguez, Emilio Rodrigo Zhuma Mera

    Technological incidents represent a critical challenge for telecommunications companies due to their direct impact on service availability, customer satisfaction, and operational continuity. This study analyzes frequent patterns of technological incidents reported by a telecommunications company in Ecuador during the peri-od 2021–2024, with the aim of supporting incident management through a data-driven approach aligned with ITIL v4. A total of 36,878 historical incident rec-ords were analyzed using data science techniques, including exploratory data analysis, normalization, value imputation, and correlation analysis. Subsequently, unsupervised data mining algorithms (Apriori, FP-Growth, and ECLAT) were applied to identify recurrent incident patterns. The results reveal significant con-centrations of incidents associated with service degradation and failures in cus-tomer-end equipment, with higher incidence in specific geographic areas. Based on these patterns, an incident management reference framework composed of eight components is proposed, adapted to the operational needs of the company and aligned with ITIL v4 practices. The proposed framework facilitates incident prioritization and decision-making in IT service management, contributing to con-tinuous improvement supported by empirical evidence.