abstract: EMBRACE (Enhanced Moved By Reading to Accelerate Comprehension in English) is an IPad application that uses the Moved By Reading strategy to help improve the reading comprehension skills of bilingual (Spanish speaking) English Language Learners (ELLs). In EMBRACE, students read the text of a story and then move images corresponding to the text that they read. According to the embodied cognition theory, this grounds reading comprehension in physical experiences and thus is more engaging. In this thesis, I used the log data from 20 students in grades 2-5 to design a skill model for a student using EMBRACE. A skill model is the set of knowledge components that a student needs to master in order to comprehend the text in EMBRACE. A good skill model will improve understanding of the mistakes students make and thus aid in the design of useful feedback for the student.. In this context, the skill model consists of vocabulary and syntax associated with the steps that students performed. I mapped each step in EMBRACE to one or more skills (vocabulary and syntax) from the model. After every step, the skill level is updated in the model. Thus, if a student answered the previous step incorrectly, the corresponding skills are decremented and if the student answered the previous question correctly, the corresponding skills are incremented, through the Bayesian Knowledge Tracing algorithm. I then correlated the students’ predicted scores (computed from their skill levels) to their posttest scores. I evaluated the students’ predicted scores (computed from their skill levels) by comparing them to their posttest scores. The two sets of scores were not highly correlated, but the results gave insights into potential improvements that could be made to the system with respect to user interaction, posttest scores and modeling algorithm. / Dissertation/Thesis / Masters Thesis Computer Science 2016

El texto completo de este trabajo no está disponible en el Repositorio Académico UPC por restricciones de la casa editorial donde ha sido publicado. / This research proposes a prediction model that might help reducing the dropout rate of university students in Peru. For this, a three-phase predictive analysis model was designed which was combined with the stages proposed by the IBM SPSS Modeler methodology. Bayesian network techniques was compared with decision trees for their level of accuracy over other algorithms in an Educational Data Mining (EDM) scenario. Data were collected from 500 undergraduate students from a private university in Lima. The results indicate that Bayesian networks behave better than decision trees based on metrics of precision, accuracy, specificity, and error rate. Particularly, the accuracy of Bayesian networks reaches 67.10% while the accuracy for decision trees is 61.92% in the training sample for iteration with 8:2 rate. On the other hand, the variables athletic person (0.30%), own house (0.21%), and high school grades (0.13%) are the ones that contribute most to the prediction model for both Bayesian networks and decision trees.

A multitude of data is being produced from the increase in instructional technology, e-learning resources, and online courses. This data could be used by educators to analyze and extract useful information which could be beneficial to both instructors and students. Educational Data Mining (EDM) extracts hidden information from data contained within the educational domain. In data mining, hybrid method is the combination of various machine learning techniques. Through this dissertation, the novel use of machine learning hybrid techniques was explored in EDM using three educational case studies. First, in consideration for the importance of students’ attention, on and off-task data to analyze the attention behavior of the students were collected. Two feature selection techniques, Principal Component Analysis and Linear Discriminant Analysis, were combined to improve the classification accuracies for classifying the students’ attention patterns. The relationship between attention and learning was also studied by calculating Pearson’s correlation coefficient and p-value. Our examination was then shifted towards academic performance as it is important to ensuring a quality education. Two different 2D- Convolutional Neural Network (CNN) models were concatenated and produced a single model to predict students’ academic performance in terms of pass and fail. Lastly, the importance of using machine learning in online learning to maintain academic integrity was considered. In this work, primarily a traditional machine learning algorithms were used to predict the cheaters in an online examination. 1D CNN architecture was then used to extract the features from our cheater dataset and the previously used machine learning model was applied on extracted features to detect the cheaters. Such type of hybrid model outperformed the original traditional machine learning model and CNN model when used alone in terms of classification accuracy. The three studies reflect the use of machine learning application in EDM. Classification accuracy is important in EDM because different educational decisions are made based on the results of our model. So, to increase the accuracies, a hybrid method was employed. Thus, through this dissertation it was successfully shown that hybrid models can be used in EDM to improve the classification accuracies.

The overall goal of this work is to support student success in computer science. First, I introduce BitFit, an ungraded practice programming tool built to provide students with a pressure-free environment to practice and build confidence working through weekly course material. BitFit was used in an introductory programming course (CSC 110) at the University of Victoria for 5 semesters in 2015 and 2016. The contributions of this work are a number of studies done analyzing the log data collected by BitFit over those years. First, I explore whether patterns can be identified in log data to differentiate successful from unsuccessful students, with a specific focus on identifying students at-risk of failure within the first few weeks of the semester. Next, I separate out only those students who struggle early in the semester, and examine their changes in programming behaviour over time. The goal behind the second study is to differentiate between transient and sustained struggling, in an attempt better understand the reasons successful students are able to overcome early struggles. Finally, I combine survey data with log data to explore whether students understand whether their study habits are likely to lead to success. Overall, this work provides insight into the factors contributing to behavioural change in an introductory programming course. I hope this information can aid educators in providing supportive intervention aimed at guiding struggling students towards more productive learning strategies. / Graduate / 0984 / 0525 / 0710 / aestey@uvic.ca

Submitted by William Justo Figueiro (williamjf) on 2015-07-28T20:32:18Z No. of bitstreams: 1 27d.pdf: 4180188 bytes, checksum: 617cda1d8cedaa910bb66676e96c33d1 (MD5) / Made available in DSpace on 2015-07-28T20:32:18Z (GMT). No. of bitstreams: 1 27d.pdf: 4180188 bytes, checksum: 617cda1d8cedaa910bb66676e96c33d1 (MD5) Previous issue date: 2014-04-15 / Nenhuma / Aplicações que fazem uso de tecnologias como Mineração de Dados Educacionais (MDE) e Learning Analytics (LA) vêm sendo adotadas na mitigação da evasão escolar, disponibilizando informações sobre os alunos que são utilizadas em intervenções pedagógicas. Os trabalhos estudados sobre a implementação destas aplicações priorizam a descrição das técnicas empregadas e existem poucas avaliações da sua utilização em larga escala, além da falta de detalhamento sobre as causas da evasão. Este trabalho apresenta um estudo de fatores envolvidos no fenô- meno de evasão escolar e descreve a utilização de um sistema para MDE e LA durante 18 meses em cursos de graduação na modalidade de Educação a Distância. É ampliada a análise dos fatores tradicionalmente monitorados e utilizados nos sistemas de MDA e LA, com a inclusão de elementos associados ao papel exercido pelos docentes e pelo conjunto de aspectos metodológicos de cada instituição. O sistema possui como diferencial a flexibilidade na integração e utilização dos dados gerados no processo de mediação digital, o que permite que necessidades de diferentes ferramentas de apoio sejam disponibilizadas. Resultados positivos destacados são a identificação de perfis de alunos evasores e a realização de intervenções pedagógicas, com redução das médias da evasão. / Educational Data mining (EDM) and Learning Analytics (LA) applications have been adopted in mitigation of dropout, providing information about students who are employed in pedagogical interventions. The most papers about the implementation of these systems describe the techniques employed, there are few evaluations of their large-scale use, apart from the lack of detail about the causes of dropout. This work presents a study of factors involved in dropout and describes the use of a system for EDM and LA during 18 months for undergraduate courses in distance education. The analysis of the factors traditionally monitored and used in EDM and LA systems is extended, with the inclusion of elements associated with the role exercised by the teachers and by institutional methodological aspects. The system has flexibility in integration and use of data generated in the process of digital mediation, which allows different support tools to be available. Some results are the identification of evaders students profiles and the realization of pedagogical actions with reducing evasion.

abstract: Due to large data resources generated by online educational applications, Educational Data Mining (EDM) has improved learning effects in different ways: Students Visualization, Recommendations for students, Students Modeling, Grouping Students, etc. A lot of programming assignments have the features like automating submissions, examining the test cases to verify the correctness, but limited studies compared different statistical techniques with latest frameworks, and interpreted models in a unified approach. In this thesis, several data mining algorithms have been applied to analyze students’ code assignment submission data from a real classroom study. The goal of this work is to explore and predict students’ performances. Multiple machine learning models and the model accuracy were evaluated based on the Shapley Additive Explanation. The Cross-Validation shows the Gradient Boosting Decision Tree has the best precision 85.93% with average 82.90%. Features like Component grade, Due Date, Submission Times have higher impact than others. Baseline model received lower precision due to lack of non-linear fitting. / Dissertation/Thesis / Masters Thesis Computer Science 2019

Submitted by Jos? Henrique Henrique (jose.neves@ufvjm.edu.br) on 2018-05-23T19:00:06Z No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) warley_leite_fernandes.pdf: 2460576 bytes, checksum: d929e82a2e47dac8f54b1a1d52ed28fb (MD5) / Approved for entry into archive by Rodrigo Martins Cruz (rodrigo.cruz@ufvjm.edu.br) on 2018-06-05T14:49:36Z (GMT) No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) warley_leite_fernandes.pdf: 2460576 bytes, checksum: d929e82a2e47dac8f54b1a1d52ed28fb (MD5) / Made available in DSpace on 2018-06-05T14:49:36Z (GMT). No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) warley_leite_fernandes.pdf: 2460576 bytes, checksum: d929e82a2e47dac8f54b1a1d52ed28fb (MD5) Previous issue date: 2017 / A Educa??o a Dist?ncia (EAD) tem-se confirmado como importante ferramenta de capacita??o a qualquer tempo e dist?ncia. Por?m, a maioria das Institui??es de Ensino tem encontrado dificuldades relacionadas ao grande n?mero de abandono dos cursos. Avan?os recentes em diversas ?reas da tecnologia possibilitaram o surgimento das Tecnologias da Informa??o e Comunica??o que se tornaram essenciais ? condu??o dos processos educacionais. Assim, imensos volumes de dados s?o gerados pela intera??o de usu?rios em Ambientes Virtuais de Aprendizagem (AVA). Esses dados ?escondem? informa??es ricas. Contudo, manipular tamanha quantidade de dados n?o ? uma tarefa simples. Neste sentido, uma solu??o promissora para extra??o de informa??o ? a Minera??o de Dados, que pode ser entendida como a transforma??o de dados brutos em conhecimento. Essa pesquisa apresenta um estudo para compreender os motivos do baixo desempenho dos alunos em cursos t?cnicos da EAD aplicando, para isto, o algoritmo de Classifica??o Associativa (CBA) em Minera??o de Dados Educacionais (EDM). Com o objetivo de gerar os melhores resultados preditivos de Classifica??o Associativa obtidos pelo CBA, aplicou-se o algoritmo de Regras de Associa??o denominado Predictive Apriori,ainda n?o empregados em trabalhos correlatos. Os resultados experimentais apontam que o CBA aplicado a Bases de Dados Educacionais atinge melhores resultados que os algoritmos de classifica??o tradicionais (alcan?ando uma marca de 85% de acur?cia). Mostrou-se tamb?m que o uso das ferramentas f?rum, quiz e folder t?m uma grande influ?ncia no desempenho dos estudantes. / Disserta??o (Mestrado Profissional) ? Programa de P?s-Gradua??o em Educa??o, Universidade Federal dos Vales do Jequitinhonha e Mucuri, 2017. / Distance Education (EAD) has been confirmed as an important training tool at any time and distance. However, most educational institutions have encountered difficulties related to the large number of dropouts. Recent advances in several areas of technology have enabled the emergence of Information and Communication Technologies that have become essential to the conduct of educational processes. Thus, immense data volumes are generated by the interaction of users in Virtual Learning Environments (AVA). These data "hide" rich information. However, handling such a large amount of data is not a simple task. In this sense, a promising solution for information extraction is Data Mining, which can be understood as the transformation of raw data into knowledge. This research presents a study to understand the reasons of the low performance of students in technical courses of the EAD applying, to this, the Association Classification (CBA) algorithm in Educational Data Mining (EDM). In order to further improve the results obtained by the CBA, the Association Rules algorithm called Predictive Apriori, not yet employed in related works, was applied in order to generate the best predictive results of Associative Classification. The experimental results point out that the CBA applied to Educational Databases achieves better results than traditional classification algorithms (reaching a mark of 85% accuracy). It was also shown that the use of the forum, quiz and folder tools have a great influence on student performance.

abstract: While predicting completion in Massive Open Online Courses (MOOCs) has been an active area of research in recent years, predicting completion in self-paced MOOCS, the fastest growing segment of open online courses, has largely been ignored. Using learning analytics and educational data mining techniques, this study examined data generated by over 4,600 individuals working in a self-paced, open enrollment college algebra MOOC over a period of eight months. Although just 4% of these students completed the course, models were developed that could predict correctly nearly 80% of the time which students would complete the course and which would not, based on each student’s first day of work in the online course. Logistic regression was used as the primary tool to predict completion and focused on variables associated with self-regulated learning (SRL) and demographic variables available from survey information gathered as students begin edX courses (the MOOC platform employed). The strongest SRL predictor was the amount of time students spent in the course on their first day. The number of math skills obtained the first day and the pace at which these skills were gained were also predictors, although pace was negatively correlated with completion. Prediction models using only SRL data obtained on the first day in the course correctly predicted course completion 70% of the time, whereas models based on first-day SRL and demographic data made correct predictions 79% of the time. / Dissertation/Thesis / Doctoral Dissertation Educational Technology 2017

Submitted by Maria Medeiros (maria.dilva1@ufcg.edu.br) on 2018-05-24T14:28:20Z No. of bitstreams: 1 ALLAN SALES DA COSTA MELO - DISSERTAÇÃO (PPGCC) 2016.pdf: 1339540 bytes, checksum: b3bbc96d990dd77ee344c4f5434a4d00 (MD5) / Made available in DSpace on 2018-05-24T14:28:20Z (GMT). No. of bitstreams: 1 ALLAN SALES DA COSTA MELO - DISSERTAÇÃO (PPGCC) 2016.pdf: 1339540 bytes, checksum: b3bbc96d990dd77ee344c4f5434a4d00 (MD5) Previous issue date: 2016 / A evasão estudantil é uma das maiores preocupações dos institutos de ensino superior brasileiros já que ela pode ser uma das causas de desperdício de recursos da Universidade. A previsão dos estudantes com alta probabilidade de evasão, assim como o entendimento das causas que os levaram a evadir, são fatores cruciais para a definição mais efetiva de ações preventivas para o problema. Nesta dissertação, o problema da detecção de evasão foi abordado como um problema de aprendizagem de máquina supervisionada. Utilizou-se uma amostra de registros acadêmicos de estudantes considerando-se todos os 76 cursos da Universidade Federal de Campina Grande com o objetivo de obter e selecionar atributos informativos para os modelos de classificação e foram criados dois tipos de modelos, um que separa os estudantes por cursos e outro que não faz distinção de cursos. Os dois modelos criados foram comparados e pôde-se concluir que não fazer distinção de alunos por curso resulta em melhores resultados que fazer distinção de alunos por curso. / Students’ dropout is a major concern of the Brazilian higher education institutions as it may cause waste of resources. The early detection of students with high probability of dropping out, as well as understanding the underlying causes, are crucial for defining more effective actions toward preventing this problem. In this paper, we cast the dropout detection problem as a supervised learning problem. We use a large sample of academic records of students across 76 courses from a public university in Brazil in order to derive and select informative features for the employed classifiers. We create two classification models that either consider the course to which the target student is formally committed or not consider it, respectively. We contrast both models and show that not considering the course leads to better results.

Submitted by Fabio Sobreira Campos da Costa (fabio.sobreira@ufpe.br) on 2017-07-11T14:35:16Z No. of bitstreams: 3 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) projeto_v26016.pdf: 1271790 bytes, checksum: f724d8523f2ffdb11ce599aff1eb8eb6 (MD5) projeto_v26016.pdf: 1271790 bytes, checksum: f724d8523f2ffdb11ce599aff1eb8eb6 (MD5) / Made available in DSpace on 2017-07-11T14:35:16Z (GMT). No. of bitstreams: 3 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) projeto_v26016.pdf: 1271790 bytes, checksum: f724d8523f2ffdb11ce599aff1eb8eb6 (MD5) projeto_v26016.pdf: 1271790 bytes, checksum: f724d8523f2ffdb11ce599aff1eb8eb6 (MD5) Previous issue date: 2016-07-08 / As Instituições Federais de Ensino Superior (IFES) possuem um importante papel no desenvolvimento social e econômico do país, contribuindo para o avanço tecnológico e cientifico e fomentando investimentos. Nesse sentido, entende-se que um melhor aproveitamento dos recursos educacionais ofertados pelas IFES contribui para a evolução da educação superior, como um todo. Uma maneira eficaz de atender esta necessidade é analisar o perfil dos estudantes ingressos e procurar prever, com antecedência, casos indesejáveis de evasão que, quanto mais cedo identificados, melhor poderão ser estudados e tratados pela administração. Neste trabalho, propõe-se a definição de uma abordagem para aplicação de técnicas diretas de Mineração de Dados objetivando a classificação dos discentes ingressos de acordo com o risco de evasão que apresentam. Como prova de conceito, a análise dos aspectos inerentes ao processo de Mineração de Dados proposto se deu por meio de experimentações conduzidas no ambiente da Universidade Federal de Pernambuco (UFPE). Para alguns dos algoritmos classificadores, foi possível obter uma acurácia de classificação de 73,9%, utilizando apenas dados socioeconômicos disponíveis quando do ingresso do discente na instituição, sem a utilização de nenhum dado dependente do histórico acadêmico. / The Brazilian's Federal Institutions of Higher Education have an important role in the social and economic development of the country, contributing to the technological and scientific advances and encouraging investments. Therefore, it is possible to infer that a better use of the educational resources offered by those institutions contributes to the evolution of higher education as a whole. An effective way to meet this need is to analyze the profile of the freshmen students and try to predict, as soon as possible, undesirable cases of dropout that when earlier identified can be examined and addressed by the institution's administration. This work propose the development of a approach for direct application of Data Mining techniques to classify newcomer students according to their dropout risk. As a viability proof, the proposed Data Mining approach was evaluated through experimentations conducted in the Federal University of Pernambuco. Some of the classification algorithms tested had an classification accuracy of 73.9% using only socioeconomic data available since the student's admission to the institution, without the use of any academic related data.