Especialización en Estadística Aplicada

URI permanente para esta colección

Examinar

Envíos recientes

Mostrando 1 - 5 de 391
  • Ítem
    Predicción de la Deforestación en Caldas, Colombia: Un Enfoque Sociodemográfico mediante el Uso de Modelos Machine Learning
    (Fundación Universitaria Los Libertadores. Sede Bogotá., ) Rivera Basto, Alejandra; Pajoy Peña, Sebastián; González Veloza, José John Fredy
    In the context of growing concerns about deforestation and its environmental and social impacts, this research focuses on understanding the relationships between sociodemographic characteristics and deforestation in the department of Caldas, Colombia. Purpose. The main purpose of this study is to develop a regression-based Machine Learning model that predicts deforestation based on sociodemographic variables, aiming to facilitate territorial planning and deforestation mitigation in the region. Methodology. The research was conducted by collecting sociodemographic data from the National Department of Statistics (DANE) and satellite records of deforestation from the University of Maryland. These datasets were integrated to analyze the relationship between population and deforestation. Machine Learning algorithms and computational tools were used to develop and evaluate regression models. Results. The results reveal significant relationships between sociodemographic variables and deforestation in Caldas. The Machine Learning model accurately predicted deforestation, providing a valuable tool for data-driven territorial planning. Conclusions. According to these results, the need to implement territorial planning policies that take into account population characteristics becomes evident. These characteristics include age, especially those above 60 years old, educational levels, considering that individuals with higher education tend to contribute less to deforestation, and the occupation of the population. These elements are highlighted as key factors in the management and mitigation of deforestation in Caldas.
  • Ítem
    Priorización de casos de hospitalización de dengue a partir de variables geográficas y de entorno usando machine learning
    (Fundación Universitaria Los Libertadores. Sede Bogotá., ) Villagrán Solorzano, Norma Alexandra; Briceño Uron, Jorge Eduardo; González Veloza, José Jhon Fredy
    Dengue is a viral disease of significant public health concern in Colombia due to its impact on numerous regions and its association with social, demographic, and environmental factors. In the year 2023, there has been an increase in the number of dengue cases compared to previous years during the same period. In terms of complexity, the World Health Organization (WHO) classifies the disease into Dengue without warning signs (Type A), Dengue with warning signs (Type B), and Severe Dengue, for which the defined care and hospitalization process outlined in the Comprehensive Clinical Care Guide for Dengue Patients is followed, based on the patient’s symptoms. Purpose. This study aims to prioritize the population of Type A and Type B dengue cases in the national territory with a higher likelihood of requiring hospitalization. This prioritization is based on sociodemographic variables, as well as climatic and geographic factors such as temperature and altitude above sea level. The goal is to provide a supplementary tool for the healthcare system to facilitate triage in patient care and treatment, thereby preventing the development of severe clinical conditions and maximizing available healthcare capacity. This tool is especially valuable during frequent dengue outbreaks.Methodology. This study utilized sociodemographic data of dengue patients treated in the national territory published in the National Public Health Surveillance System (Sivigila), Code 210, for the years 2019, 2020, 2021, and 2022 (n= 315076). Additionally, it incorporated climatic and geographic conditions from the municipalities where cases occurred. Several supervised machine learning classification models were trained using this information. Results. The model with the best result on test data was the Extreme Gradient Boosting (xgboost), which yielded an AUC of 67%. The highest hospitalization rate is observed in patients between 11 and 8 years old, from low socioeconomic strata, and with a symptom period exceeding 4 days. Regarding geographical location, an average altitude of 235 m above sea level and an average temperature of 27.8°C favor hospitalization conditions, as well as the notification of infection between weeks 1 to 6 and 46 to 52 of the year.
  • Ítem
    Análisis Estadístico del homicidio en Colombia: Una vista enfocada a los perfiles sociodemográficos, educativos y económicos
    (Fundación Universitaria Los Libertadores. Sede Bogotá., ) Cabarcas Mondol, Cabarcas Mondol; Castañeda Correa, Mónica; Van Strahlen Martínez, Esmeralda; Pineda Ríos, Wilmer
    This research focuses on the analysis of statistical data on homicides in Colombia between the years 2016 and 2020. Purpose. The main objective is to examine the sociodemographic, educational and economic profiles of the population in order to identify patterns and trends that explain the violent acts that trigger homicides. Methodology. This research addressed the high homicide rate in Colombia, by analyzing the rate by department and exploring relevant variables. A data mining methodology, including statistical methods and machine learning techniques, was extracted to analyze the collected data (11 variables, 1981 observations) using CRISP DM methodology. The information for the analysis was obtained from the pages of the National Police, DANE, ECLAC, DNP (National Planning Department) and HUMANITARIAN DATE EXCHANGE. Results. In the analyzes carried out, it is not possible to identify in a representative way a variable of those analyzed that increases the homicide rate, an analysis must be carried out that groups more variables. conclusions. This study suggests that education, poverty, and unemployment may play an important role in determining the homicide rate in the region studied. However, further analysis and consideration of other factors is needed to fully understand the dynamics and underlying causes of homicides in the region. These evidences determined an initial basis for future research that addresses the problem of violence and improves socioeconomic conditions in the region.
  • Ítem
    Pronóstico de calificación de la asignatura “Morfofisiología 1” de la Facultad de Salud de una Universidad de Bogotá utilizando un modelo SARIMA.
    (Fundación Universitaria Los Libertadores. Sede Bogotá., ) Caranton Pineda, Felipe Antonio; Niño, Carolina
    In this article, a statistical model is used to predict grades for a health school subject, pretending to be an initial input to improve the academic and educational process of its teaching. Purpose. Determine the best statistical model of SARIMA time series to better predict the qualifications of the subject Morphophysiology 1 of the Faculty of Health of a university in Bogotá. Methodology. A univariate SARIMA model was used that describes the autocorrelations of the data, establishing the most appropriate forecast for the data set. Results. The forecast of the ratings for a horizon of 24 months with ARIMA (1,1,1) showed an average value of 4.17. Conclusions. For the next 24 months, grades close to 4.2 would be expected, which indicates that future students will be successful in passing this subject if the pedagogical teaching model is continued.
  • Ítem
    Análisis estadístico de pruebas educativas en física: identificación de patrones y recomendaciones
    (Fundación Universitaria Los Libertadores. Sede Bogotá., ) Gonzáles Limas, José Mauricio; Saenz Lesmez, Ross Mary; Romero Ospina, Manuel Francisco
    The research consisted of carrying out the evaluation of the educational tests of four Physics subjects of the Los Libertadores University Foundation for the years 2019-2020. For this, statistical tools for grouping and main components were used, from the TRI reliability assessment to the identification of behavior patterns of individuals and thus generate a series of results that related the discrimination index, the grouping of learning by subject and the individuals to carry out a qualitative reflection on the design of the questions to the extent that they must answer why the general domain does not correspond to the particularity of each test.