Yujing Zhou

Managed various kinds of API methods (GET/POST/PATCH) in Python and Flask; built the bridge to deliver back-end work to be displayed on front-end applications. Created DAG running recurrent tasks via Airflow to support daily database updates. Automated property valuation process by implementing complex strategy to suggest a final valuation, based on different cluster analysis. Used Apache Spark (in Python) to cost-effectively calculate area statistics/valuations of properties to minimize risk on acquisition.

Processed raw data from insurance company. Built algorithms in R to remove duplicated claims (5% in over 40,000 claims); extracted statistical interpretation to provide insights to external customers. Researched on machine learning model to support executive ideas for more robust data driven decisions. Used SMOTE oversampling algorithm to deal with imbalanced data problem, increasing F1 score of over 80% features. Used ROC curve to select features and increase the recall of Naive Bayes classifier from 90% to 95%.

Worked on Multi-stage Decisions Modeling, building new R packages - rewrote the program of consideration set model(MATLAB) in R and test it on real data; did splines and change point exploration based on simulated data.

Worked on data processing and generated transaction tables with Python; processed the exchange of stock index futures data by linear regression.

Worked on Successful Cases Modeling with SQL & R. Used random forest algorithm to find common features among customers who purchased the service of securities margin trading.