Cici L. Email & Phone Number

Bozeman, Montana, Us

Developing optimized strategy using FICO Decision Optimizer

Bozeman, Montana, Us

• Leading optimization projects and delivering decision tree models for credit line and price optimization• Leading model development and validation for credit risk models and payment projection models • Deploying and testing analytic models into FICO Blaze Advisor, TRIAD, and FICO Cloud platforms• Managing client projects, conducting client training, supervising junior analyst’s work

Bozeman, Montana, Us

• Financial credit risk modeling at account, product, and customer level development. Building origination and account management models for risk products for banks in North and South Americas. This includes the analysis of large amounts of historical data, data cleaning and filtering, roll rate analysis, sampling, exclusions and performance definition, variable generation, variable selection, experiments with different types of algorithms and models, analyzing performance to identify the best algorithms to employ.• Developed machine learning models (SGB, FSSGB, random forest) to predict insurance fraud.• Built decision tree model for credit line and price management• Deployed analytic models into FICO Blaze Advisor, TRIAD, and FICO Cloud platforms. • Developed provision models for client to predict lifetime Expected Credit Loss (manipulating 550 billion rows in Hadoop) ) to comply with IFRS 9 accounting standard, which included building PD model, Forward Look model for PD, EAD, and Stage Transition. Deployed production code in HiveQL.• Developed Indirect Price Elasticity model for Auto Loan portfolio using FICO Analytic Modeler Scorecard Pro.• Developed Early Payment Default models to predict the first party fraud for an appliance retailer that issues loans to consumers.

New York, Ny, Us

• Developed multi-dimensional segmentations to differentiate customer’s distinct purchase choice, channel preference, and engagement level with client brand using Latent Class Segmentation• Built Email & Direct Mail Engagement Indices that ranked order customer's engagement level with different communication channel• Presented multiple successful analytical projects to client

New York, Ny, Us

• Developed propensity models that drove brand conversion, external prospecting, and seasonal campaign response• Built nonlinear mixed model that fit both baseline response and incrementality for an eyeglass retailer and was used in direct marketing to maximize campaign incrementality• Laid out intricate experimental design for monthly mailing campaigns, optimized targeted offers given constraints

New York, Ny, Us

• Performed SQL query to extract campaign leads for marketing campaign in Teradata • Developed a segmentation typing tool that expanded segmentation capability using recursive partitioning tree in R• Automated dashboard reporting to track KPI, customer base monitoring, performed ad-hoc analysis

New York, Ny, Us

Developed response models (logistic regression) for Marketing Campaigns; Tracked model performance and campaign KPI; Evaluated data quality across multiple demographic data vendors.

• “Research on Intelligent Forecasting Technique based on Support Vector Regression” supported by National Natural Science Foundation of China, supervised by Prof. Yukun Bao• Collected macroeconomic time series data and compared prediction accuracy of traditional Time Series Forecasting methods (ARIMA, Simple Exponential Smoothing, Holt-Winters) with modern Machine Learning techniques (SVM, Neural Networks etc)• Proposed a hybrid modeling framework and compared with straightforward use of SVM, ARIMA, EEMD-ARIMA-SUM, EEMD-ARIMA-SVM approaches in forecasting.

Master, Operations Research

Bachelor, Management Information System