Yasmine El Gharoussi

– Applied unsupervised clustering (K-Means, GMM, DBSCAN) on high-dimensional datasets to extract latent behavioral patterns, methodology directly transferable to market segmentation and customer profiling – Designed rigorous statistical benchmarking frameworks (supervised vs. unsupervised; Random Forest vs. XGBoost) with uncertainty quantification and interpretability, following econometric evaluation principles – Built reproducible end-to-end Python pipelines (data cleaning, feature engineering, statistical modeling, evaluation) with structured reporting for stakeholder communication – Presented findings in English at international conferences (IMES Cambridge, Mecano); co-authored peer-reviewed publication (arXiv, 2026), synthesizing complex analyses into clear strategic narratives

– Cleaned, structured and analyzed behavioral and subscription data of SOHO/SME segments using PySpark and SQL to identify engagement patterns, default risk, and commercial growth levers across the full customer base – Developed and validated propensity models (Random Forest, XGBoost) end-to-end, from raw data ingestion to model evaluation and deployment-ready output, supporting strategic decisions on subscription plan management – Designed Power BI dashboards to track key commercial KPIs in real time, enabling actionable visibility for business and operations teams – Delivered client-ready recommendations to cross-functional stakeholders, translating model outputs into strategic decisions and clearly communicating findings to non-technical audiences