Algorithms and programming are some of the most challenging topics faced by students during undergraduate programs. Dropout and failure rates in courses involving such topics are usually high, which has raised attention towards the development of strategies to attenuate this situation. Machine learning techniques can help in this direction by providing models able to detect at-risk students earlier. Therefore, lecturers, tutors or staff can pedagogically try to mitigate this problem. To early predict at-risk students in introductory programming courses, we present a comparative study aiming to find the best combination of datasets (set of variables) and classification algorithms. The data collected from Moodle was used to generate 13 distinct datasets based on different aspects of student interactions (cognitive presence, social presence and teaching presence) inside the virtual environment. Results show there are no statistically significant difference among models generated from the different datasets and that the counts of interactions together with derived attributes are sufficient for the task. The performances of the models varied for each semester, with the best of them able to detect students at-risk in the first week of the course with AUC ROC from 0.7 to 0.9. Moreover, the use of SMOTE to balance the datasets did not improve the performance of the models.

The present work describes the challenges faced during the development of a countrywide Learning Analytics study and tool focused on tracking and understanding the trajectories of Uruguayan students during their first three years of secondary education. Due to the large scale of the project, which covers an entire national educational system, several challenges and constraints were faced during its conception and development. Examples of key challenges were: to understand the different nuances of the educational system, to satisfy ethical and legal requirements without narrowing down the scope and potential novelty of the initiative and to deal with integration and inconsistencies in the databases. This paper presents the design decisions and solutions found to address or mitigate the problems found as a contribution to facilitate similar large-scale projects. Three main experiments using data mining were conducted and their results are also described, pointing out the feasibility of finding meaningful patterns that can be used by educational authorities to improve their decision making and foster public educational policies. Finally, the paper describes the use of the data and the findings of the study to create an interactive visual tool to explore the relationship between student variables, performance and persistence in the system.

Papanicolaou is an inexpensive and non-invasive method, generally applied to detect cervical cancer, that can also be useful to detect cancer on oral cavities. Although oral cancer is considered a global health issue with 350.000 people diagnosed over a year it can successfully be treated if diagnosed at early stages. The manual process of analyzing cells to detect abnormalities is time-consuming and subject to variations in perceptions from different professionals. To evaluate a possible solution to the automation of this process, in this paper we employ the object detection deep learning approach in the analysis of this type of image using 3 models: RetinaNet, Faster R-CNN, and Mask R-CNN. We trained and tested the models using images from 6 cytology slides (4 cancer cases and 2 healthy samples) and our results show that Mask R-CNN was the best model for localization and classification of nuclei with an IoU of 0.51 and recall of abnormal nuclei of 0.67.