Mengmeng Liu Email & Phone Number

Rockville, Md, Us

Rockville, Maryland, United States

Greater Atlanta Area

I worked as a data scientist intern in the Model Risk and Management team in Equifax. • Designed and implemented one sensitivity analysis method to evaluate the sensitivity of correlated explanatory variables in various models (e.g., Logistic Regression, DNN, CNN) based on real big data in Python.• Conducted R&D on Explanatory AI (XAI), try to explain the DNN models from different aspects (e.g., sensitivity, SHAP).May 2019 - Aug 2019: data scientist intern in Fraud Detection team within Data & Analytics Group• Created 210 new fraud attributes based on experts’ inputs and domain knowledge from multiple data sets by HiveQL / Impala queries in Hue and Bash.• Designed advanced machine learning models to improve Synthetic ID and First Party Fraud detection in Python (Skope-Rule & XGBoost): New fraud attributes improve the fraud capture rate by 7% (reduce fraud loss by $8+ Millions/month).• Conducted R&D on CNN for automatic attributes creation. o Created 100 deep attributes by CNN using time-series data. o New CNN attributes improve the fraud capture rate by 19% (reduce fraud loss by $3.3 Millions/month).

Atlanta

GTA for OMSA course "Data and Visual Analytics"

Greater Atlanta Area

• Analyzed the sustainability of moving interstates within I-285 underground and compared it with the sustainability of the current situation. I used twelve sustainability indicators from economic, environmental and social dimension to measure connectivity, commute time, energy cost, air pollution, land use, accessibility (to green space, health facilities, and education facilities), and equity. I analyzed the spatial correlations between these twelve indicators using R packages. I build the composite sustainability index for each dimension of sustainability by a linear additive model. I analyzed the spatial distributions of each indicator and each composite sustainability index in QGIS.

• Designed and graded homework assignments and class project for the course “Data and Visual Analytics". • Hold office hours to answer students’ questions

Greater Atlanta Area

• Designed and developed a spatial sustainability assessment framework for civil infrastructures in my doctoral research. It is a data-driven framework with the support of spatial database (PostgreSQL). This framework provides a systematical way to evaluate the sustainability of civil infrastructures, including indicator selection, multi-criteria decision analysis, uncertainty and sensitivity analysis, and multiple visualizations of sustainability evaluation results.• Evaluated the vulnerability of the transportation network in Atlanta, by examining the change of accessibility to health facilities after moving the top ten edges with the largest edge-betweenness one by one.• Analyzed the human resource data of GDOT and created maps to show the information in ArcGIS and on the website.• Analyzes and predicted PM2.5 in Beijing from 2009 to 2015 using three regression models. They are Autoregressive Moving Average Model, Neural Network, and Support Vector Regression. The first model is good at short-term (within 50 days) prediction, while the latter two are better at the long-term forecast.• Constructed food flow network using 2012 US Commodity Flow Survey data, and optimized the network by minimizing its total “food miles” by Linear Programming. The optimized US food flow network has the potential to reduce “food miles” by more than 50%, which result in a smaller sustainability impact by reducing transportation GHS emissions.

Doctor Of Philosophy - Phd, Computational Science And Engineering, 3.69/4.0

Master Of Science - Ms, Computational Science, 3.80/4.0

Master Of Science - Ms, Geographic Information Science And Cartography, 4.36/5.0

Bachelor Of Science - Bs, Geographic Information Science And Cartography, 4.56/5.0