There is a wide variety of statistical tools developed for making decisions in the business conte... more There is a wide variety of statistical tools developed for making decisions in the business context. The analysis of functional data is a area of study of growing importance in the last years. In the present paper some of its applied techniques are proposed to make analysis of financial risk. Specifically, results from exploratory functional analysis, identification of atypical data and the construction of supervised classification models based on the risk classification of credit unions, subject to the control of the Superintendency of Banks of Ecuador, taking as functional variables the NPL ratio and the net profit margin in the period July 2011 to December 2012.
This article presents the application of a methodological procedure for the construction of a sta... more This article presents the application of a methodological procedure for the construction of a statistical qualification model for the approval of commercial credits in a public financial institution. In this line, the main aim is to reveal the benefits of using generalized additive models (GAM), whose functional structures contemplate the possible non-linearity of the explanatory variables of credit risk in relation to compliance with the payment obligations of borrowers, compared to linear models like the logit. This topic becomes relevant in view of the need for financial institutions to have the right tools and information management systems that allow them to de-establish strategies to improve the placement of their loan portfolio with clients who can fulfill their agreed obligations within the established deadlines, without incurring partial or total delays; in short, minimizing your credit risk. Additionally, in order to meet the stated need, the methodological procedure is ap...
Cybersecurity attacks are considered among the top five of risks worldwide, according to the Worl... more Cybersecurity attacks are considered among the top five of risks worldwide, according to the World Economic Forum in the year 2019. This context has generated the need to improve the tasks of cybersecurity defense in organizations. Improving the effectiveness in executing a cybersecurity task requires three pillars: people, processes and technologies. The proposal in this work is to analyze the integration of these three components as a strategy to improve the effectiveness of the execution of operational tasks in cyber defense, specifically the detection of anomalies. Based on the foundation that: cybersecurity operational tasks carried out daily by analysts require the use of cognitive processes, and that the use of techniques based on technologies such as machine learning, data mining and data science have generally been used to automate cybersecurity tasks, we have considered the use of cognitive security, as a strategy to improve the anomaly detection process, taking into account the cognitive processes and skills that are executed by the security analyst.
Resumen-En el presente artículo se aplica un modelo de aprendizaje automático supervisado que pre... more Resumen-En el presente artículo se aplica un modelo de aprendizaje automático supervisado que predice la probabilidad de que un estudiante de la Escuela Politécnica Nacional apruebe el curso de nivelación. Para llevar a cabo esta tarea se describe una metodología estadística basada en gradient boosting y regresión logística donde el problema de aprendizaje se formula en términos de la minimización de la función de error mediante el método del descenso del gradiente. Para explicar la probabilidad de aprobación se toman en consideración dimensiones sugeridas por la literatura relacionadas a variables socioeconómicas, demográficas, familiares, institucionales y de desempeño académico en la postulación y en el curso de nivelación que tiene el estudiante. Los resultados del modelo de árbol de decisión muestran un nivel de precisión del 96% en el conjunto de datos de prueba, con un área bajo la curva ROC de 89.1, siendo estos niveles generalmente aceptados. Por otro lado, los resultados de la regresión logística sugieren que factores como la calificación ponderada del primer bimestre, la calificación con la que postuló, su jornada de estudios, su ubicación geográfica de origen, entre otras, afectan de una u otra manera a la probabilidad del estudiante, de aprobar el curso de nivelación. Palabras clave-rendimiento académico, regresión logística, árboles de decisión, GBM, método del descenso del gradiente.
This paper investigates the existence of spatial regimes of high violence levels across Mexican m... more This paper investigates the existence of spatial regimes of high violence levels across Mexican municipalities. Our approach consists of providing a framework to explicitly address spatial heterogeneity, which might suggest instability in the structural determinants of homicides. In this context, a distinction is made in relation to the regimes in municipalities within states with long-standing trafficking activities by comparing those municipalities that have been exposed to joint operations (operativos conjuntos) and those that were not exposed to the operations. Spatial econometric models were estimated for each regime to investigate possible spillover effects arising from the covariates. The results point to differences in regard to the significance, magnitude, and sign of the effects related to some variables according to each spatial regime's specification. While the direct effects show that socioeconomic variables tend to play an important role in explaining the variation...
Integration of Administrative Records for Social Protection Policies-A Systematic Literature Review of Cases in the Latin American and Caribbean Region
Trends in Artificial Intelligence and Computer Engineering, 2022
Proceedings of the 8th International Conference on Data Science, Technology and Applications, 2019
The aim of this work is to propose different statistical and machine learning methodologies for i... more The aim of this work is to propose different statistical and machine learning methodologies for identifying anomalies and control the quality of energy efficiency and hygrothermal comfort in buildings. Companies focused on energy sector for buildings are interested on statistical and machine learning tools to automate the control of energy consumption and ensure quality of Heat Ventilation and Air Conditioning (HVAC) installations. Consequently, a methodology based on the application of the Local Correlation Integral (LOCI) anomaly detection technique has been proposed. In addition, the most critical variables for anomaly detection are identified by using ReliefF method. Once vectors of critical variables are obtained, multivariate and univariate control charts can be applied to control the quality of HVAC installations (consumption, thermal comfort). In order to test the proposed methodology, the companies involved in this project have provided the case study of a store of a clothing brand located in a shopping center in Panama. It is important to note that this is a controlled case study for which all the anomalies have been previously identified by maintenance personnel. Moreover, as an alternatively solution, in addition to machine learning and multivariate techniques, new nonparametric control charts for functional data based on data depth have been proposed and applied to curves of daily energy consumption in HVAC.
Field-based fire studies in the equatorial Andes indicate that fires are strongly associated with... more Field-based fire studies in the equatorial Andes indicate that fires are strongly associated with biophysical and anthropogenic variables. However, fire controls and fire regimes at the regional scale remain undocumented. Therefore, this paper describes spatial and temporal burned-area patterns, identifies biophysical and anthropogenic fire drivers, and quantifies fire probability across 6° of latitude and 3° of longitude in the equatorial Andes. The spatial and temporal burned-area analysis was carried out based on 18 years (2001-2018) of the MCD64A1 MODIS burned-area product. Climate, topography, vegetation, and anthropogenic variables were integrated in a logistic regression model to identify the significance of explanatory variables and determine fire occurrence probability. A total of 5779 fire events were registered during the 18 years of this study, located primarily along the western cordillera of the Andes and spreading from North to South. Eighty-eight percent of these fires took place within two fire hotspots located in the northwestern and southwestern corners of the study area. Ninety-nine percent occurred during the second part of the year, between June and December. The largest density of fires was primarily located on herbaceous vegetation and shrublands. Results show that mean monthly temperature, precipitation and NDVI during the prefire season, the location of land cover classes such as forest and agriculture, distance to roads and urban areas, slope, and aspect were the most important determinants of spatial and temporal fire distribution. The logistic regression model achieved a good accuracy in predicting fire probability (80%). Probability was higher in the southwestern and northern corners of the study area, and lower towards the north in the western and eastern piedmonts of the Andes. This analysis contributes to the understanding Spatiotemporal patterns of burned areas, fire drivers, and fire probability across the equatorial Andes
Violence has increased all around Mexico in the last years, reflecting an uprise in the rate of h... more Violence has increased all around Mexico in the last years, reflecting an uprise in the rate of homicides, and especially after some federal intervention took place to fight the drug cartels in some states. In this paper we use data at the municipal level to link social and institutional factors with the rates of homicides. We exploit the entrance for federal army interventions in 2007 and 2008 in some states to fight drug cartels. Using different estimation methods, we find that inequality, access to social security and income, as well as local provision of security and law are relevant in explaining homicides. We also find that the army interventions have increased not only drug related homicides, but also general homicides in municipalities under intervention compared with those with no intervention.
Miguel Flores, Ph.D. (c) en Estadistica e Investigacion de Operaciones y M.A. en Tecnicas Estadis... more Miguel Flores, Ph.D. (c) en Estadistica e Investigacion de Operaciones y M.A. en Tecnicas Estadisticas por la Universidad de La Coruna, Espana; Eduardo Marin, M.A. en Estadistica Aplicada con mencion en Econometria y Data Mining por la Universidad Catolica de Lovaina, Belgica; Victor Morales, Dr. (c) en Estadistica, Universidad de Valparaiso, Chile, y estudiante de la Maestria de Investigacion en Economia del Desarrollo de FLACSO Ecuador
Skills Developed by Economics Students During Their Professional Training Year
Despite the increasing improvement in student degree qualifications, there is evidence that gradu... more Despite the increasing improvement in student degree qualifications, there is evidence that graduates lack “employability” skills. The Professional Training Year (PTY) programme at the University of Surrey contributes to closing the gap between degree qualifications and graduate employability skills since students on placements tend to acquire not only specific job skills but also transferable skills. This research project has been conducted by an academic staff member and a student working in partnership. It aims to identify the skills required in Economics placements through document analysis based on the information from Economics students who completed their PTY during 2017–2018. This chapter will thus provide a better understanding of what skills are required within Economics placements and how these learning approaches might help promote wider employability skills development.
Esta investigación desarrolla un modelo de la dinámica de los rendimientos del Índice de Precios... more Esta investigación desarrolla un modelo de la dinámica de los rendimientos del Índice de Precios y Cotizaciones (IPC) de la Bolsa Mexicana de Valores (BMV) en el marco de la teoría de valores extremos y en particular lleva a cabo una aplicación de la distribución generalizada de Pareto. El análisis estadístico muestra que muchas de las observaciones son inusuales (grandes en valor absoluto) y que no pertenecen al mundo de la distribución Normal. De hecho se muestra que dichas observaciones superan sistemáticamente un umbral. El objetivo principal es mostrar que el enfoque de valores extremos con picos proporciona una descripción más adecuada de los rendimientos del IPC que los modelos que solamente emplean el supuesto de normalidad.
El presente trabajo de investigación surge en relación con las necesidades de la industria petrol... more El presente trabajo de investigación surge en relación con las necesidades de la industria petrolera para producir aquellos yacimientos del Distrito Morichal, poseedores de altas reservas remanentes, donde MEOR simboliza una alternativa biotecnológica, efectiva y rentable que emplea la inyección de microorganismos como una estrategia adicional a los tratamientos convencionales de recuperación mejorada de hidrocarburos. Distintos campos petroleros situados en diversas partes del mundo han implementado este proceso con enfoques y expectativas similares, sin embargo, los resultados han tenido discrepancias, puesto que las bacterias se ven afectadas significativamente por las variaciones en las condiciones del subsuelo, es por ello, que se diseñó una metodología de selección de pozos candidatos para llevar a cabo esta técnica, donde se exhibe el ambiente más idóneo para tales fines, aunado a ello, se establecieron ciertas correlaciones a partir de programas estadísticos para predecir ta...
Uploads
Papers by Miguel Flores