Curricular analysis aims to gain insight into the strengths and weaknesses of a curricular plan, identifying problems such as deviations in students' trajectories and bottlenecks, as well as evaluating the flexibility and constraints of the paths. Educational Process Mining (EPM) ofers a process-centric perspective on curricular data. A key challenge for EPM is the availability of tools for curricular analysis that non-technical users can interpret, are open-source, and are readily available for use. In this paper, we present the Trajecta tool, which provides a model-driven approach for curricular analysis using Process Mining, benefiting both students and academic managers in managing curricular data. The tool ofers a web interface, a Java component with a generic metamodel for modeling curricular programs, and a microservices Python component that leverages the PM4Py library to provide process mining functionalities. It was evaluated using an example application with real data from the Computer Science program at the Facultad de Ingeniería (FING), Universidad de la República (UdelaR).
Value Trajecta 1.2 GPL-3.0 Metadata description Tool name Current version Legal code license Download/Demo URL Documentation URL Source code repository Screencast video Languages, tools and services used Java, Python, Ecore, PM4Py, JavaScript, HTML, Flask, Servlets, Supported operating environment https://gitlab.fing.edu.uy/open-coal/trajecta.git https://gitlab.fing.edu.uy/open-coal/trajecta https://gitlab.fing.edu.uy/open-coal/trajecta https://gitlab.fing.edu.uy/open-coal/trajecta/-/raw/main/demo.mp4 CEUR
ceur-ws.org (D. Calegari)
Curricular analysis [
In [
To close this gap, we propose the Trajecta tool, which supports curricular analysis for both students and academic managers, helping them manage curricular data. The Trajecta tool ofers a model-driven approach for modeling curricular programs. It is based on a metamodel that defines common yet generic concepts and relations for key elements, such as plans and courses, which can be used to model diferent careers from various institutions. It provides key functionalities, such as curricular trajectory path recommendations for students, and the application of key process mining analysis techniques, including discovering students’ trajectory models, analyzing deviations, identifying bottlenecks, and generating data statistics for academic managers. The tool was evaluated using real data for the Computer Science career at the Facultad de Ingeniería (UdelaR), validating its support for curricular modeling and analysis.
The rest of the document is organized as follows: Section 2 describes the functionalities of the tool, Section 3 introduces the architecture of the tool and technologies used, Section 4 discussed the maturity of the tool and finally Section 5 presents some conclusions and future work.
The tool follows a model-driven approach, which, at its core, is a metamodel for curricular programs depicted in Figure 1, allowing the modeling of a wide range of study programs. It integrates key concepts, such as institutions and their careers, along with concrete study plans for a career, i.e., a set of partially ordered curricular units that a student must complete. Moreover, it supports more dynamic concepts, such as instances of curricular units (i.e., courses over a given period), various types of periodic evaluations (partial, course, and exam), and students’ enrollments in courses and assessments, along with their corresponding grades. A curricular plan is defined by creating a model that complies with the metamodel, which serves as the reference model for the curricular plan and is used for subsequent analysis and evaluation.
The tool provides two main categories of functionalities, each one designed for a specific user: i) curricular trajectory recommendation, which aims for students that want to know the available paths they can follow at any given moment, and ii) curricular trajectory analysis, which aims for academic managers that want to evaluate how students are traversing the curricula defined, where are the deviations, the bottlenecks, and diferent statics. In the following, these two categories and the main functionalities provided are introduced, illustrating the tool’s use with an application example using real data from the Computer Science career at the Faculty of Engineering (UdelaR).
Students can model their academic trajectory by entering information about their career goals, course plans, and evaluation results, which are then visualized as a graph. Figure 2a presents an example of a trajectory for a student of the Computer Science career at the Faculty of Engineering (UdelaR). The tool provides recommendations on curricular units that the student can enroll in, using a color-coded scheme: green for fully approved, yellow for partially approved, and white for available curricular units for enrollment.
(a) Example student trajectory (b) Example career plan curricula
Students can access detailed information about curricular units, including prerequisites, associated subjects (e.g., mathematics), and other attributes. The curricula can be visualized as a directed graph, where arrows represent prerequisite relationships. For example, Figure 2b illustrates that enrollment in “Programación 2” (Programming 2), of the Computer Science career at the Faculty of Engineering (UdelaR), requires prior approval of “Programación 1” (Programming 1). This supports students in planning their studies, understanding the requirements for specific courses, and identifying available options at each stage of their career.
For academic managers, Trajecta integrates process mining techniques through the PM4Py [
The tool currently supports: 1. Discovery of students’ academic trajectories, 2. Statistical analysis of curricular units and curricular plans, 3. Visualization of performance graphs and trajectory graphs in multiple formats, 4. Evaluation of students’ performance over time, 5. Conformance checking for deviations between student trajectories and curricular plans.
Although these are commonly used process mining perspectives for curricular analysis, the user interface presents its functionalities in educational terminology to be easily understood by domain people. Through these capabilities, academic managers gain insight into how students progress, where deviations occur, and how curricular structures influence academic performance. Figure 3a depicts an example of Petri net discovery of students’ trajectories, showing other discovery options provided by the tool, and Figure 3b presents an example of deviations showing alignments between variants of student trajectories with respect to the curricular plan (reference model). (a) Discovered Petri net from the computer science (b) Alignment graph between students from 2018 students’ event log to 2021 to the curricular plan
The tool is presented as a web application, where the front-end can be deployed on any static
server or service, such as GitHub Pages, or accessed directly from disk by opening the .html
ifles. It uses vis.js to visualize the trajectory of a student and the curriculum of a career plan.
It employs choices.js to enable searching within the select inputs of the tool. The backend is
divided into microservices: the process mining ones are written in Python and use PM4Py [
The Java application provides functionalities for curricula trajectory recommendation, including a metamodel with EMF support. The Python microservices support the process mining functionalities for the category of curricular trajectory analysis. The Java component also converts a student trajectory that is written in EMF to an event log. The Python component, designed with microservices, allows for easy extension of functionalities.
The tool has been evaluated using real data from the Computer Science program at the Faculty of Engineering (UdelaR). The model-driven approach was validated through the modeling of the curricular plan and the use of its functionalities by university students (with access to the front-end component and Java functionalities)2. The process mining component, i.e., the PM4Py microservices backend, was validated using real data from the computer science career, specifically students’ trajectories from 2018 to 2021. The event logs contain 2,297 cases (students) and 27,920 events, along with 1,708 variants, spanning from February 7, 2018, to March 5, 2021. Some performance limitations were identified during log discovery and graph visualization, such as the discoveries being executed in a background thread and not notifying 1https://eclipse.dev/emf/ 2https://trajecta.fapret.com/ the user when they are finished. Although initial tests were conducted primarily by students for the model-driven part, we plan to extend it for the process mining analysis component.
We have presented the Trajecta tool, which provides a model-driven approach for curricular analysis with Process Mining. The tool is open-source and available for free use. It provides support for non-technical users and facilitates the modeling and analysis of diferent careers, serving as a basis for comparative analysis. We are integrating additional process mining functionalities to support specific educational analyses, focusing on more user-friendly visualizations than those already provided. We are also evaluating the adaptability of the metamodel to model various types of study programs at diferent universities. This will also provide the basis for comparative process mining analysis between diferent institutions and countries.
During the preparation of this work, the authors used Grammarly to check grammar and spelling, paraphrase, and reword. After using them, the authors reviewed and edited the content as needed, taking full responsibility for the publication’s content.
We thank the Facultad de Ingeniería (FING), Universidad de la República (UdelaR) for providing the academic data and the Computer Engineering career curricula used in the demonstration. We also thank Santiago Freire, who contributed to the initial MDE-based development.