lalaproject@cti.espol.edu.ec

Programa

Seleccione el día que desee revisar

Lunes y Martes: Conferencia y Workshop

Lunes

Martes

Hora Lunes
08:30 – 09:00 Acreditación – Auditorio Edificio 9000
09:00 – 09:30 Bienvenida Autoridades – Auditorio Edificio 9000 Keynote – Auditorio Edificio 9000

Tinne De Laet
Dashboard de Aprendizaje Orientados a Estudiantes: El (sin)
sentido de las posibilidades de éxito y los modelos predictivos.
09:30 – 10:30 Keynote – Auditorio Edificio 9000

Carlos Alario
Toma de decisiones a partir del análisis de datos educativos.
10:30 – 11:00 CAFÉ
11:00 – 11:30 Presentaciones

Auditorio Edificio 9000 Sala 9201
Scaling Learning Analytics up to
the national level: the experience from Estonia and Uruguay
Learning Analytics, dashboard for
academic trajectory
11:30 – 12:00
Predicting academic results in a
modular computer programming course for
engineering freshmen
Analysis of data in an educational
platform with MMOG features
12:00 – 12:30
Describing educational trajectories
of engineering students in individual high-failure rate
courses that lead to late dropout
Learning Analytics en Colombia: una
revisión de la literatura y un análisis del esfuerzo de
investigación en el ámbito local
12:30 – 13:00
Professional networks in online learning
processes
A learning analytics experience using
interaction visualization dashboards to support virtual
tutoring
10:00 – 10:30 Presentaciones

Auditorio Edificio 9000 Sala 9201
Meta-Predictive Retention Risk
Modeling: Risk Model Readiness Assessment at Scale
with X-Ray Learning Analytics
Effort analysis of computational thinking
process over a gamified and non-gamified environments
10:30 – 11:00
Learning Analytics in Mathematics
Teacher Education at the Ceará State University
Analyzing the influence of online
behaviors and learnings approaches on academic
performance in first year engineering
11:00 – 11:30
Traceability of learning activities in
computer programming courses using an automatic
online judge within an LMS
Towards an automatic text-based
emotion classifier for supporting emotional learning in
a higher education institution
11:30 – 13:30 Workshops

Auditorio Edificio 9000 Sala 9201
LALA Framework – Parte 1: De la necesidad
institucional a las Analíticas de Aprendizaje en el aula.
Using the ENA Webtool
13:00 – 14:00 Almuerzo – Sala 9101
13:30 – 14:30 Almuerzo – Sala 9101
14:30 – 16:30 Workshops

Auditorio Edificio 9000 Sala 9201
LALA Framework – Parte 2: Metodología Iterativa paso a paso
para diseñar herramientas de analíticas de aprendizaje
Reflexionando sobre mi práctica docente con TrAC
16:30 – 17:30 Cierre Evento
14:00 – 15:00 Posters
15:00 – 17:00 Workshops

Auditorio Edificio 9000 Sala 9201
Utilización de R para analítica
del aprendizaje
Diseño de Visualizaciones para
ambientes de enseñanza hibrida: retos y oportunidades para los docentes
18:00 – 22:00 Cena – Restaurant Cocina del Bosque en Punucapa

Punto de encuentro para la salida sector El Péndulo de la costanera   Transporte en Catamarán Tornagaleones a las 18 horas.  Regreso desde Punucapa es a las 22 horas

 

 

Resumenes Charlas

 

Toma de decisiones a partir del análisis de datos educativos

Tradicionalmente los procesos de enseñanza y aprendizaje se han desarrollado tomando como referencia la propia experiencia de profesores y alumnos, muchas veces sin tener en cuenta los avances derivados de investigaciones previas en el ámbito educativo. Sin embargo, hoy en día los datos son esenciales para la toma de decisiones en educación. Los profesores deben apoyarse en los datos para mejorar los contenidos que ofrecen a los alumnos y rediseñar sus clases. Los estudiantes deben apoyarse en los datos para mejorar sus habilidades de autorregulación del aprendizaje. El potencial de un uso correcto de los datos afecta a cualquier nivel educativo (educación preuniversitaria, educación superior, aprendizaje a lo largo de la vida), y contexto educativo (educación presencial, educación virtual, educación híbrida – blended).

 

Dashboard de Aprendizaje Orientados a Estudiantes:

El (sin) sentido de las posibilidades de éxito y los modelos predictivos

¿Los dashboard de aprendizaje son soluciones escalables y sostenibles para proporcionar retroalimentación útil a los estudiantes? ¿Analíticas de aprendizaje son aplicable en entornos de educación superior tradicionales?. En esta charla se compartirán experiencias y lecciones aprendidas de dos proyectos europeos (ABLE and STELA) que tuvieron como propósito el desarrollo de dashboard de aprendizaje para instituciones de educación superior tradicionales e integrarlos en prácticas educativas concretas. La charla desafiará sus creencias con respecto a las «posibilidades de éxito» y el uso de modelos predictivos.

 

Regresar

 

Workshops

 

Utilización de R para analítica del aprendizaje

Este workshop proporciona conocimiento para aplicar algunas metodologías estadísticas y técnicas de análisis de datos a través de la herramienta R en procesos educativos. Se introduce la herramienta estadística R, algunos casos de estudio en el ámbito del aprendizaje con datos educativos y algunas técnicas de aplicación de analítica del aprendizaje. Se prestará especial atención a técnicas de predicción (por ejemplo, redes bayesianas, regresión lineal o árboles de decisión) y con especial énfasis en casos de predicción de abandono.

Universidades: Universidad Carlos III de Madrid y Universidad de
Edimburgo

Presentadores: Pedro J. Muñoz Merino, Carlos Alario Hoyos, Carlos
Delgado Kloos, Jon Imaz Marín,
Nora’Ayu Ahmad Uzir, Wannisa Match

Idioma: Inglés

Prerequisitos: Programación Básica, Instalar R en sus computadores
personales y llevarlos al taller

Duración: 2 horas

Regresar

 

 

Diseño de Visualizaciones para ambientes de enseñanza hibrida: retos y
oportunidades para los docentes

Este taller busca identificar y proponer posibles visualizaciones que contribuyan a apoyar las actividades de enseñanza y aprendizaje en ambientes de aprendizaje híbridos (blended). Durante el taller se analizarán las necesidades de 42 profesores de 10 países con experiencia en este tipo de contextos. Posteriormente, los participantes trabajarán evaluando varias visualizaciones usadas en herramientas previamente propuestas para apoyar a los profesores en otros contextos. Finalmente, todos los participantes trabajarán en el diseño de un tablero de visualización que servirá de base para el desarrollo de sus propias herramientas y de la herramienta NoteMyProgress.

Universidades: Pontificia Universidad Católica de Chile y Universidad
Católica de Lovaina (KU Leuven)

Presentadores: Ronald Perez, Jorge Maldonado y Tom Broos

Idioma: Español

Prerequisitos: Sin Prerequisitos

Duración: 2,5 horas

Material: https://drive.google.com/open?id=1AwWamp-rw4WwE2iGfImeoo2czK31xem0

Regresar

 

 

Using the ENA Webtool

The workgroup will conduct a hands-on session on the Epistemic Network Analysis (ENA) webtool. ENA is an educational network analysis that visualises the structure of connections between concepts and measures the strength of association among these concepts in a network. It also quantifies changes in the composition and the strengths of connections over time. In this workshop, we’ll demonstrate how to use ENA to analyse educational data, such as conversations and engagement patterns in the virtual learning environment. Learners will learn to use the ENA Web tool and R to analyse data using the ENA method.
.

Universidades: Universidad de Edimburgo

Presentadores: Rafael Ferreira, Nora’Ayu Ahmad Uzir, Wannisa Match y
Jane Sinclair

Idioma: Inglés

Prerequisitos: Computadores personales con acceso a Internet

Duración: 2 horas

Regresar

 

 

LALA Framework – Parte 1: De la necesidad institucional a las Analíticas de Aprendizaje
en el aula.

En los últimos años, la análitica de aprendizajes o Learning Analytics (LA) ha captado la atención de los líderes de educación superior que vieron en esta línea de investigación una fuente de información para la toma de decisiones institucionales. Sin embargo, la mayor parte de la literatura actual en LA se ha concentrado en el desarrollo de herramientas y métodos para apoyar la actividad a pequeña escala, sin abordar necesariamente cómo podría adoptarse a gran escala. Se han propuesto varios marcos teóricos para guiar la adopción de LA a nivel institucional, pero faltan trabajos empíricos para ilustrar las implicaciones de la adopción de LA en la vida real. Para guiar la adopción de LA en instituciones de educación superior en Latinoamérica, este taller presenta el marco LALA: un enfoque multidimensional basado en la experiencia de cuatro universidades de la región. Específicamente, este taller abordará la dimensión institucional de la adopción de LA mediante el uso del LALA Canvas, una plantilla que tiene como objetivo guiar una discusión grupal sobre el estado actual de una institución de educación superior en términos del uso de datos educativos. Varios desafíos políticos, técnicos y éticos serán discutidos durante el taller, de manera que los participantes anticipen las implicancias del diseño y la implementación de futuras herramientas de LA en sus institucione

Universidades: Pontificia Universidad Católica de Chile, Universidad
Carlos III de Madrid y Universidad de Edimburgo

Presentadores: Isabel Hilliger, Jorge Maldonado, Pedro J. Muñoz Merino
y Miguel Angel Zuñiga

Idioma: Español

Prerequisitos: Sin Prerequisitos

Duración: 2 horas

Regresar

 

 

Reflexionando sobre mi práctica docente con TrAC

Este taller busca presentar la herramienta TrAC para profesores, ésta herramienta ha sido creada mediante sesiones de codiseño con docentes de la Universidad Austral de Chile con el objetivo de que el uso de ella fomente la reflexión de los docentes en su práctica, basándose en indicadores creados a partir de datos académicos históricos de sus asignaturas. Durante el taller se expondrá a través de escenarios uso de la herramienta diseñada y los participantes podrán utilizarla, comentar, reflexionar y proponer mejoras a la propuesta para facilitar la adopción en sus propios contextos educativos.

Universidades: Universidad Austral de Chile

Presentadores: IJulio Guerra, Eliana Scheihing, Cristian Olivares y
Valeria Henriquez

Idioma: Español

Prerequisitos: Computadores personales con acceso a Internet

Duración: 2 horas

Regresar

 

 

LALA Framework – Parte 2: Metodología Iterativa paso a paso para diseñar herramientas de analíticas de
aprendizaje

Un punto clave en el proceso de diseñar una herramienta que contenga analíticas de aprendizaje, es pasar desde la obtención de requerimientos al diseño de la misma. Debido a esto, muchas interrogantes aparecen en la mente de los diseñadores, investigadores y programadores: ¿Qué hacer una vez que se obtienen los requerimientos? ¿Cómo diseño la herramienta? ¿Qué aspectos técnicos debo considerar? ¿Qué preguntas formulo en la etapa de pruebas? En este taller, trabajaremos a través de un caso de estudio real, las diferentes etapas que conllevan al diseño de una herramienta, empezando por el levantamiento de requerimientos, su organización, pasando por la etapa de pruebas con prototipos de baja fidelidad, hasta la iteración con el usuario y producción.

Universidades: Escuela Superior Politécnica del Litoral (ESPOL)

Presentadores: Margarita Ortiz, Alberto Jimenez, Ricardo Maya y Pedro
Lucas

Idioma: Español

Prerequisitos: Computadores personales con acceso a Internet

Duración: 2 horas

Regresar

 

 

Papers Aceptados

 

 
  

Scaling Learning Analytics up to the national level: the experience from Estoniaand Uruguay

Autores: Adolfo Ruiz Calleja, Sofía García, Kairit Tammets, Cecilia Aguerrebere and Tobias Ley
Tipo: Conferencia
Presenta: Mg. Sofía García Cabeza, Fundación Ceibal – Flacso Uruguay, sofiagarciacabeza@gmail.com


Abstract

This paper analyzes the key aspects in the implementation of national-level LA and the limitations that current initiatives present. With this purpose, we present a multiple case study that describes six nationallevel LA projects in Uruguay and Estonia. By means of a data value chain, we synthesize the steps followed by the LA projects to extract meaning out of data and discusses the main issues related to scaling up LA. We found out that these LA projects are driven by political – and not so much by educational- aspects. We also saw that integrating personal data from different educational institutions is a key step in these projects, which entails technical, legal and administrative issues. We also see a big potential in national-level LA to provide a macro perspective that supports teachers, students, parents or school leaders to take evidencebased decisions.

 

Regresar

 

 
  

Predicting academic results in a modular computer programming course for engineering freshment

Autores: Claudio Alvarez, Alyssa Wise, Sebastian Altermatt and Ignacio Aránguiz
Tipo: Conferencia
Presenta: Claudio Alvarez, Universidad de los Andes,Chile, calvarez@uandes.c


Abstract

At present, computer programming skills are essential in engineering curricula and professional practice. In spite of this, and after decades of research in programming pedagogy, academic success in introductory programming courses continues to be a challenge for many students. In this research we explore the feasibility of predicting academic results in a modular computer programming course in a Chilean university (N=242), through measurement of psychometric variables linked to implicit theories of intelligence, error orientation, and students’ attitudes towards programming. Coincidentally with other recent studies conducted in Finland and Turkey, early measurement of implicit theories of intelligence did not emerge as a predictor of academic performance in the programming course. As for error orientation, students exhibiting mild measures of an error strain construct did seem to perform better than students with extreme measures. The variables with the highest predictive potential were found to be students’ attitudes towards programming; namely, their perceived value of programming skills, and perception of programming self-efficacy. Substantial differences were noted in both latter constructs among male and female students. We discuss implications of our findings and future research prospects

 

Regresar

 

 
  

Describing educational trajectories of engineering students in individual high-failure rate courses that lead to late dropout

Autores: Juan Pablo Salazar-Fernández, Marcos Sepúlveda and Jorge Muñoz-Gama
Tipo: Conferencia
Presenta: Juan Pablo Salazar, Universidad Austral de Chile, juansalazar@uach.cl


Abstract

Understanding the phenomenon of late dropout has been gaining importance in engineering education. The process that leads to the decision to drop an academic program after freshman year, usually covers several academic periods, and is influenced by different factors. Longitudinal analysis is a good approach to analyze previous events that lead to late dropout. In this case study, we use a process mining approach to answer how educational trajectories of engineering students may describe the process that finishes in late dropout. It was conducted at the Universidad Austral de Chile, using academic records of high-failure rate courses. Through the analysis of educational trajectories in each one of the high-failure rate courses, we found that trajectories of students that dropout early and those that graduate on-time are clearly distinguishable from each other, but the trajectories of students that dropout late or graduate late need to be analyzed in more detail. Late dropout is higher among students that fail in freshman year courses compared to those that fail in sophomore year courses, and passing a course after failing it several times can be a dropout risk factor, which is consistent with the Investment Model. Such findings may be useful for managers and policy makers, because these trajectories can be related to entrance conditions and permanence requirements

 

Regresar

 

 
  

Professional networks in online learning processes

Autores: Sofía García Cabeza, Antonio López Arredondo, Pablo Massaferro, Nicolás Rubido and Alvaro Margolis Hirt
Tipo: Conferencia
Presenta: Mg. Sofía García Cabeza, Fundación Ceibal – Flacso Uruguay, sofiagarciacabeza@gmail.com


Abstract

In the present study, an online continuing medical education course is analyzed as it evolves. We investigate the benefits of usingSocial Network Analysis and Educational Data Mining techniques to predict andimprove the students’ performance, as well as to tailor the course into suitable groups of students and teachers, in order to enhance the course results. Specifically, we use information from participants, shared prior to the course, as the motor to create an acquaintance network, which evolves as the course evolves. These social links between participants allow us to characterize their interactions in relationship to the course, hence, establishing the role they play in their social network, for example, as hubs or outliers. In particular, this allows us to group participants according to their social closeness. Moreover, using Educational Data Mining techniques, we can forecast the students’ potential performance in the course and design instructional strategies accordingly. This research shows the importance of working in multidisciplinary teams to address social problems, and also, to bring forth to education research knowledge and techniques that have been successfully used in other research fields.

 

Regresar

 

 
  

Learning Analytics, dashboard for academic trajectory

Autores: Mario Patricio Peña Ortega, Fabian Bravo and Lourdes Illescas Peña
Tipo: Conferencia
Presenta: Lourdes Illescas Peña, Universidad de Cuenca, lourdes.illescasp@ucuenca.edu.ec


Abstract

In University academic management context, several proposals for analysis and visualization learning trajectories studies have been developed. Taking account that the educational trajectory is the student path traveled at a given time from income to completion of the stay, then it can be considered that the use of technology would either extract or highlight relevant information that is not directly visible with the traditional tools. The data visualization at educational environments has become into a challenge due large amounts of available information. The responsibility of educational managers demand a clear visual proposal and adapted to queries based on an academic context. Therefore, it was proposed to generate a dynamic visualization tool based on relevant variables of the students. To do this, the proposal began with a review of the literature that helped analyze the different ways of visualizing the data of academic trajectories. Afterward, a dynamic visualization was formulated in order to explain teachers and authorities through learning analysis dashboard, based on use of parallel coordinates that present multidimensional data over time scale. The sample was constituted by 1975 students records of an Ecuadorian university, of the cohort that began on March 2013, distributed by faculties and careers. The used technique allowed us finding out trends and relationships among dimensions, improving the student trajectory patterns understanding, dropout trends, either increase or decrease performance, among other relationships. The queries allowed filter data by variables such as: faculties, careers, students and score intervals. Finally, the proposal validation was performed based on the dashboard relevance, in response of academic manager queries.

 

Regresar

 

 
  

Analysis of data in an educational platform with MMOG features

Autores: Néstor Darío Duque-Méndez, Emilcy J Hernández-Leal and Julian Moreno Cadavid
Tipo: Conferencia
Presenta: Emilcy J. Hernández, Universidad Nacional de Colombia, ejhernandezl@unal.edu.co


Abstract

Erudito is an online authoring tool to create and monitor MMOG-type education-al games (Massively Multiplayer Online Game) aimed at interactively re-creating the teaching and learning process in a virtual classroom, in a challenging, collabo-rative, and funny way. The learning activities in Erudito generate a large amount of data at different levels, related to processes carried out by students. Being so, the aim of this paper is to present the application of different techniques of data analysis, taking advantage of the records obtained from the students ‘interaction in Erudito, in order to reflect the trends and patterns of students’ performance with respect to their behavior in the learning activities carried out.

 

Regresar

 

 
  

Learning Analytics en Colombia: una revisión de la literatura y un análisis del esfuerzo de investigación en el ámbito local.

Autores: Oscar Eduardo Cala Wilches and Victor Hugo Grisales Palacio
Tipo: Conferencia
Presenta: Oscar Cala, Universidad Nacional de Colombia, oscarcala@gmail.com


Abstract

Recientemente el campo de estudio conocido como Learning Analytics o analíticas de Aprendizaje ha tenido un importante desarrollo. Este desarrollo se debe en gran medida al esfuerzo de investigadores que, en su mayoría, se encuentran ubicados en Estados Unidos y Europa. En América Latina se han realizado varios esfuerzos para poder consolidar una comunidad de trabajo alrededor de este campo de estudio. Con el objetivo de apoyar dichos esfuerzos, fortalecer la comunidad investigativa en América Latina y consolidar la presencia de los investigadores colombianos en dicha comunidad, este documento presenta una revisión exhaustiva de los trabajos realizados en Colombia, relacionados con las analíticas del aprendizaje. Se presenta una revisión cualitativa, cuantitativa y narrativa, que permite identificar las instituciones e investigadores más relevantes durante los últimos años, así como discutir algunos cuestionamientos relacionados con los niveles de adopción en Colombia y los posibles beneficios que podrían obtenerse si se logra incrementar la aplicación de las analíticas de aprendizaje en el sistema educativo colombiano.

 

Regresar

 

 
  

A learning analytics experience using interaction visualization dashboards to support virtual tutoring.

Autores: Luis Magdiel Oliva Córdova, Héctor R. Amado-Salvatierra, Leonel Monterroso Torres, Maylin Suleny Bojórquez Roque and Klinge Villalba Condori.
Tipo: Conferencia
Presenta: Leonel Monterroso, Secretaría Nacional de Ciencia y Tecnología de Guatemala, lmonterroso@senacyt.gob.gt


Abstract

We live in the digital age of data, and the technologies we use allow us to store different records that can serve as inputs for decision making in any learning environment. This work was developed in a virtual university context, applying learning analytics based on interaction visualization boards. The study was based on the question: how to support tutoring in a virtual course through the use of visualization boards of student activity? The research method used was formulated under a quasi-experimental cross-sectional design, with a mixed approach, developed through an introductory course on mobile technologies and virtual platforms for learning. The study involved 55 students who had no previous experience of training in virtual environments. The total number of participants was divided into groups and moderated by 5 virtual tutors at the beginner level. The results present an experience of Learning Analytics (LA) using visualization dashboards to support the work of the virtual tutor, highlighting that through graphics, it is easier to identify the most active students within a group, verify access to tasks and URL resources, visualize the number of submissions of an activity, establish percentages of dedication, evaluate performance through qualifications and intervene in a didactic and pedagogical way when necessary

 

Regresar

 

 
  

Meta-Predictive Retention Risk Modeling: Risk Model Readiness Assessment at Scale with X-Ray Learning Analytics

Autores: Aleksander Dietrichson, John Whitmer and Diego Forteza.
Tipo: Conferencia
Presenta: John Whitmer, ACT, Inc., john.whitmer@act.org


Abstract

Deploying X-Ray Learning Analytics (Blackboard Inc, 2015) at scale presented the challenge of deploying customized retention risk models to a host of new clients. Prior findings made the researchers believe that it was necessary to create customized risk models for each institution, but this was a challenge to do with the limited resources at their disposal. It quickly became clear that usage patterns detected in the Learning Management System (LMS) were predictive of the later success of the risk model deployments. This paper describes how a meta-predictive model to assess clients’ readiness for a retention risk model deployment was developed. The application of this model avoids deployment where not appropriate. It is also shown how significance tests applied to density distributions can be used in order to automate this assessment. A case study is presented with data from two current clients to demonstrate the methodology.

 

Regresar

 

 

 
  

Learning Analytics in MathematicsTeacher Education at the Ceará State University

Autores: João Batista Carvalho Nunes, Viviani Maria Barbosa Sales and João Bosco Chaves.
Tipo: Conferencia
Presenta: João Batista Nunes, Universidade Estadual do Ceará, joao.nunes@uece.br


Abstract

Enrollment in higher education is growing in Brazil. Teacher education, particularly, faces several problems pointed out by studies in the country. Low completion rates of undergraduate teacher education programs requires urgent measures to change this situation with quality. Learning analytics emerges as an option to improve students’ learning, reducing the risk of academic failure. This research, as a part of a larger project, sought to develop a statistical model that helps to predict students at risk of not being successful in the subjects of the Undergraduate Mathematics Teacher Education Program, in distance education modality, at the Open University of Brazil / Ceará State University (UAB/UECE). For that, a statistical method was used. It was decided to use data of the students who entered the university in 2009, in poles of two cities in the countryside of Ceará: Mauriti and Piquet Carneiro. The students’ access and interaction data in the Moodle environment and final performance grades in the subjects were collected. In data analysis, binary logistic regression was used. The predictive model developed has the academic situation of the student (approved / failed) as a dependent variable and, as independent variables, the student’s action categories in Moodle. This model has six independent variables, related to the activities: choice, forum, questionnaire and task, and to the file resource. It has a high overall efficiency, with statistically significant predictor variables and high accuracy percentages to predict failure and non-failure.

 

Regresar

 

 
  

Traceability of learning activities in computer programming courses using an automatic online judge within an LMS

Autores: Andrés Felipe Pineda Corcho and Julián Moreno Cadavid.
Tipo: Conferencia
Presenta: Andrés Felipe Pineda, Universidad Nacional de Colombia, sede Medellín afpinedac@unal.edu.co


Abstract

As many teachers may know, the only way of actually learning computer programming is through practice. Online judges may help in this task because students can practice as much as they want and receive automatic feedback about their codes correctness. Teachers, on the other side, may spend less time checking codes, and more on tutoring and advising. When embedded into an LMS, those tools might become even more helpful because the data they provide can be used for assessment purposes but also to feed learning analytics processes. Considering this scenario, we propose in this paper a process mining approach to use those data and identify student behavioral patterns. In particular, we used the Disco tool to identify interaction patterns using codes submission data as well as LMS resources access records. In order to validate our proposal, a case study was conducted in a data structures course with 101 engineering students at the National University of Colombia, in Medellín. We found well-differentiated patterns, from those that exhibit a sequential or incremental pattern, to those that reveal trial-error tactics.

 

Regresar

 

 

 
  

Effort analysis of computational thinking process over a gamified and non-gamified environments

Autores: Dante Saavedra-Sánchez, Martha Vidal-Sepúlveda and Cristian Olivares-Rodríguez
Tipo: Conferencia
Presenta: Dante Saavedra-Sanchez, Universidad Austral de Chile, dante.saavedra@alumnos.uach.cl


Abstract

In the recent years, the computational thinking has been established as a formative priority from primary school onwards, both internationally and nationally level, which has led to initiatives of recreational strategies and curricular integration. The initial training processes in programming are mainly focused on workshops about teaching methodologies, therefore, the assessment instruments are artifacts that measure the quality of the solution elaborated by the teacher, relegating the evaluation of the process due to the difficulty to obtain elements that allow the characterization. Thus, this article characterizes through analytics and statistical tests the process of solving computer problems by primary school students based on the effort involved in each challenge. Hence, experimental tests were conducted with a group of K-12 students, who solved a series of drills using a block-based tool, both gamified and non-gamified with the purpose of corroborating that gamification generates higher levels of global effort. In the end, the results show that a group of students exhibits a higher level of engagement and effectiveness in the gamified version while in the nongamified version they are more reflective. Not only do these results provide crucial information for the creation of assessment instruments of the computer solving problems process, but also provide to the behavioral differences in gamified environments

 

Regresar

 

 

 
  

Analyzing the influence of online behaviors and learnings approaches on academic performance in first year engineering

Autores: Sergio Celis, Dany López and Javier Silva.
Tipo: Conferencia
Presenta: Dany López, Universidad Católica de Chile, danylopfiqui@gmail.com


Abstract

Over the last four decades, the study of academic performance in higher education has increasingly included more sources of information to understand phenomena such as achievement or dropout. The first econometric models in the field commonly used student characteristics, pre-college achievement, and college performance. Then, a large range of psychosocial theories, with its respective instruments (typically questionnaires), added a new layer of analyses that complemented previous models. Recently, colleges and universities have dramatically expanded their capacity to capture student data through different systems. Such is the case of the learning management systems (LMS), which provide dynamic and a large amount of data about student online behavior. We are just beginning to explore how these layers of data come together to explain academic performance. In this study, we seek to understand and model these layers of data from a first year cohort at a large engineering school in Chile (784 students). First, we use support vector regressions to model second semester gpa on student characteristics, pre-college data, first semester grades, and online behavior. We then added to the model information extracted from the LEARN+ questionnaire, a psychosocial instrument that profiles different learning approaches (i.e., surface, strategic, and deep) and environmental perceptions. The results indicate that both online behavior and LEARN+ data increase prediction power. In addition to first semester performance, the features that seem to explaining academic achievement in the second semester to a significant extent are the LMS interaction distribution over the semester, perception of applied knowledge, and the score in the science score in the national admission test. These results are important for first year engineering, since in this field first year performance has long lasting effects on future persistence and achievement.

 

Regresar

 

 

 
  

Towards an automatic text-based emotion classifier for supporting emotional learning in a higher education institution

Autores: Sebastián Moreno Rodríguez, Jossie Esteban Murcia and Triviño and María Irma Díaz Rozo.
Tipo: Conferencia
Presenta: Sebastián Moreno Rodríguez, Jossie Esteban Murcia, Escuela Colombiana de Ingeniería Julio Garavito, jossie.murcia@hotmail.com, sebastian.mr47@gmail.com


Abstract

Several works have proposed text-based emotion classification models. However, Spanish language approach has not been studied enough. Generally, no study consider the context and demography drawbacks, implicit in the text. Automatic classification of basic emotions considered on the International Survey on Emotion Antecedents and Reactions (ISEAR) dataset were treated (trained and validated) with an own designed ensemble classifier schema. SemEval 2007 was used to validate our results with other investigations. The ensemble schema is based on the assumption of combining a batch of statistical machine learning classification models (NaïveBayes, Support Vector Machines and Neural networks models) and a knowledge-based tool (NRC Word-Emotion Association Lexicon), aiming to benefit from their competence to better classify certain emotions drawbacks. In order to determine the inconveniences that may arise due to a dependency on a language and a context, were proposed two approaches; ISEAR dataset translated into Spanish and our own Academic Survey on Emotional Reactions in Computational Learning Processes (ASERCLP). The analysis performed with ISEAR translated shows similar results compared to the best-trained model with the original dataset (an overall F-measure value of 62,9% vs 66,4%; respectively). The deterministic voting schema behaves in a similar way to the best results of each model. The results affirm that the context of the problem affects directly the model results. The translated datasets can be useful in the same Spanish context.

 

Regresar