Xiaotian CHEN

Implemented a modeling pipeline using PCA, clustering, DA, CART, bagging and random forests with SMOTE to predict customer churn and evaluated robustness via cross-validation.

Designed a relational database in Microsoft Access, built OLAP cubes with SAP Business Objects, wrote SQL for ETL, and created interactive Power BI dashboards for sales analysis and reporting.

Used R to clean and visualize panel data and applied fixed effects models with interactions and lags to analyze how political instability affects tourism, identifying significant negative effects and regional heterogeneity.

Developed LSTM- and BERT-based classifiers to predict mental health status from text and applied LIME for local interpretability of model predictions.

Applied g-computation and bootstrap robustness testing on synthetic data to estimate Average Causal Effects and benchmarked Lasso, Boosted CART, and neural networks to optimize causal analysis and predictive performance.

Built a CatBoost purchase prediction model on 10M+ records using Spark, achieving 50% recall via time-series feature engineering and oversampling; developed dynamic scoring with SQL window functions and delivered visualized analytics.

Master 2 in Mathématiques et Applications with specialization in ingénierie mathématique et biostatistique, including coursework in machine learning, deep learning, stochastic algorithms, survival analysis, differential models and nonparametric statistics.

Completed a Master of Econometrics and Statistics (Data Science for Social Sciences) with coursework in R, Julia, Python, SQL, Power BI, Stata, time series, Markov chains and machine learning.

Grade: 15.16/20

Bachelor of Economics covering macroeconomics, microeconomics, econometrics and statistics with practical experience in R, Python, Excel and SAS; achieved average grade 15.16/20.