PhD in Quantitative MarketingResearch interests: Targeting and Personalization, Customer Relationship Management, Influencer and Creator Economy

- Processed TB-sized electronic-invoice data by utilizing a multi-processing module in Python to enhance computational efficiency.- Implemented LDA and Word2Vec via Python to extract product representation from electronic-invoice data by taking each receipts as a document, each product as a token, and applied clustering algorithms to observe consumers' purchasing behavior in different industry.- Led a team of research assistants to label 20K+ products, and applied supervised machine learning techniques to classify products into categories.- Constructed a consumer/digital price index CPI/DPI) using electronic-invoice data by Laspreyes formula.- Extracted transaction record of anti-PM2.5 masks from electronic-invoice data via SQL and constructed a daily district-level market share data.- Proposed a random utility model to illustrate consumers’ preference toward anti-PM2.5 masks and applied a standard logit model derived from it to estimate consumers’ willingness to pay (WTP) for clean air.- Investigated the heterogeneity of consumers’ WTP for clean air in different regions and seasons, and provided policy implications on environmental regulation.- Conducted an independent research project on the propagating effect of social media on anti-minority incidents using Facebook Data.

- Developed a customer churn prediction model using random forest and convolutional neural network which achieved 86% of accuracy and 90% of recall rate to identify customers with a higher risk of churning.- Developed a ML-based inventory-examination system on application software for customer to check the predicted movements of their inventory stocks.- Co-designed a data-driven new-customer activation strategy by analyzing customer behavior data to identify critical time points for activation, and conducted A/B tests to evaluate campaign performance.