xgboost

Our study focused on using the Big Five personality inventory to predict traits from students' smartphone sensor data collected over 2 months under the Horizon Europe project. Through correlation analyses and machine learning with cross-validation, we showed that predictions are reliable and accurate enough for practical use.

Real-time stock market prediction system featuring a complete Big Data streaming pipeline (Kafka & Spark), machine learning models (XGBoost/Scikit-Learn), a FastAPI backend, and an interactive React dashboard.

Machine Learning case study including an exploratory data analysis and fitting a Decision Tree, Random Forest, and XGBoost Model. Interactive notebook with outputs and visualizations: https://yaldan.github.io/ml_case_study/

Early-warning machine learning system to predict mass public school closures (≥10% of schools in a district) five years in advance using NCES administrative data (2000–2018). Built to support equity-focused planning and proactive decision-making in U.S. school districts.

Deteccion remota de minas antipersona enterradas con imagenes multiespectrales (MicaSense RedEdge-MX) - trabajo de grado, EasyEnsemble XGBoost + LOSO, AUC 0.947

A robust and explainable multiclass intrusion detection system using XGBoost and TreeSHAP, featuring multi-seed evaluation, per-alert explainability, and severity-aware alert scoring for SOC-oriented cybersecurity.

Bachelor's thesis (UE Germany) — comparing Naive Bayes, Logistic Regression, SVM, Random Forest, and XGBoost for predicting diseases from patient-reported symptoms. Includes LaTeX manuscript and Python ML pipeline.

Predicting school reading proficiency from public library funding, staffing, and demographics using Lasso, Ridge, Random Forest, and XGBoost (test R² = 0.74).

Ethereum price prediction using Internal and External Factors

ML for emergency department triage: MIMIC-IV-ED evaluation, SHAP, threshold optimisation, bias audit

Leakage-aware evaluation of composition-based superconductor Tc models: code, deterministic chemical-family splits, featurized data, and figures for the MLST submission.

Código-fonte da dissertação "Detecção de Falhas em Sensores de Qualidade do Ar utilizando Aprendizado de Máquina".