ECOTEC_Online student risk dataset: An open institutional dataset for reproducible student risk analytics in online...

The ECOTEC_Online Student Risk Dataset contains anonymized institutional records from online higher education programs at Universidad ECOTEC, Ecuador. The dataset comprises 12,632 student-level records, with exactly one record per unique student, and 18 public variables covering demographic, academic, admission, financial, and institutional information relevant to student risk and institutional analytics in digital learning environments. Universidad ECOTEC provided the research team with an institutionally anonymized dataset prepared in accordance with its approved institutional protocol. The public repository is organized around the finalized anonymized dataset and includes variable-level metadata, data-validation and descriptive-analysis scripts, reproducible Logistic Regression and Random Forest benchmark scripts and outputs, supplementary visualizations, software-environment specifications, and FAIR-oriented repository documentation. TARGET_RISK is a binary variable derived during study-level data curation from restricted STUDENT_TYPE and GRADUATION_DATE information. Operationally, Class 1 denotes students who were not recorded as Alumni and had no valid GRADUATION_DATE recorded in the institutional snapshot; it should not be interpreted as confirmed educational risk or as a prospectively observed dropout outcome. It represents a contemporaneous administrative proxy based on recorded completion status rather than a prospectively observed dropout or retention outcome. Potential reuse scenarios include learning analytics, educational data mining, institutional analytics, class-imbalanced classification, explainable artificial intelligence, machine-learning benchmarking, educational decision-support research, responsible AI, and exploratory fairness-aware analyses using the demographic and institutional subgroup variables retained after anonymization. The dataset and associated reproducibility resources are publicly archived in Zenodo as version 1.0.4 (DOI: 10.5281/zenodo.22015963) [1].