Yi Zhou Email and Phone Number

•Analyze unstructured data, define key metrics, develop machine learning techniques and building data-driven dashboards to derive insights from advertising data•Conduct large-scale statistical A/B testing and machine learning model to evaluate the efficacy of advertising campaigns that serve hundreds of millions of impressions daily.•Closely work with Product team to breakdown complex business questions into actionable items, define KPIs to measure the success and interpret advertising model results into actionable insights•Collaborate with engineering team to identify correct data source, optimize code efficiency and implement automated model pipeline which can measure performance of ads campaigns

•Analyze unstructured data and build automated pipeline and Tableau dashboards to monitor KPI •Provide Offsite ads data visibility on inventory, performance, and pricing by combining products on different platforms outside of Walmart•Conduct exploratory data analysis and identify underlying patterns to generate actionable insights, such as shift budget between Offsite and Onsite Display to improve product profit

• Responsible for product health of Homepage and Personalization domains, conduct deep dives, design and evaluate a/b tests.• Successfully launched shorter POV feature to decrease bounce rate by 2% and drove 20bps lift in GMV• Define metrics for tracking the product performance and maintain effective Tableau Dashboard• Closely work with Product and Engineering partners to build and test various features to improve the performance of homepage and personalization modules and enhance the customer experience• Analyze user behavior using telemetry data to detect underlying trends in order to recommend website improvements to the product team• Create and improve beacons tracking to effectively capture user experience across Web and AppMachine Learning Projects:• Apply machine learning models on payment processing data to choose a more suitable processor based on users’ purchase history in order to improve auth success rate• Tune best parameters for Logistic Regression, KNN and Random Forest models by comparing accuracy and truth positive• launch of an intelligent payment routing project that optimizes current transaction routing logic with payment providers and helps business achieve 20bps cost savings

Achievement:· Work with product manager and engineering team to develop business intelligence dashboards.· Improve user click through rate for 5% by building movie recommendation system.· Implement A/B test to validate different data models with the engineering team.Responsibilities:Business Intelligence Dashboard:· Build Tableau dashboards to visualize insight relationship between clients’ requests and product performance. · Develop SQL queries to collect, process and clean user media usage data through Google BigQuery database.· Communicate with the engineering team and product manager to interpret analysis into meaningful suggestions.Movie Recommendation:· Use k-means clustering to assign similar clients with same labels based on their usage behavior.· Apply Classification Tree, Ada Boosting and Logistics regression methods to construct the classification model to predict new users’ type by Python.· Apply Collaborative Filtering method to find the most similar users based on users’ viewing history, and use TF-IDF to find the most similar movies based on movies’ plot summary.

·Manipulated useful transportation-related big data by SQL and inputted these table into GIS software· Implemented Machine Learning methods -- hierarchical clustering to differentiate transportation modes· Presented statistical analysis and results to technical and non-technical audiences

· Managed large data of 40,000 pairs of questions to classify whether question pairs are duplicates. · Find distance between each distinct word to constructed PPMI matrix and stored it in python sparse matrix. · Reduce the dimension of PPMI matrix by Eigen-decomposition to form word embedding matrix. · Computed the cosine similarity of two sentences and increase the efficiency from one day to two hours.

· Teaching Assistant for Elementary Statistics, responsible for 50 students· Held office hours and led discussion once a week for 10 weeks· Provided homework solutions and graded exams

· Used machine learning methods to predict if valuable employees of a big company will leave or stay.· Chose best parameters for Logistic Regression, SVM with rbf kernel, KNN, and Random Forest methods.· Compared these methods by drawing confusion matrix, ROC and PR curves using numpy, pandas and scikit-learn.

· Received investment management principles and stock analysis training· Practiced stock skills and knowledge in real-world trading platforms· Worked with colleague to analyze and present on current financial event· Learned stock market dynamics, understood investment analysis, and engaged in financial economic research

· Used Matlab to analyze the interaction of aging, movement and survival of cherry flies· Reviewed statistical outcomes, charts and graphs for Entomology department; prepared new graphs and analysis· Applied hazard function and principal component analysis on various data to reveal hidden information

· Used R and Excel to collect, execute and analyzed voters and registers’ data· Created information tables of voters and registers of different counties inCalifornia· Helped to engage public dialogue on representative governance in California

· Provided free tax return to low-income individuals and families· Used IRS forms and computer software to complete tax returns· Assisted patrons completing the intake forms and helped site coordinators with administrative tasks

• Use communication skills to assist customers.• Classified customers’ problems and guide them to different bank tellers. • Helped customers pay fees, print receipts and deposit or withdraw money. • Remember daily foreign currency exchange rates.• Gained basic knowledge of stock and fund from bank manger.

• Analyze efficient and fast spline-backfitted kernel smoothing of additive models. • Perform analysis and programming on FCAR model. • Workout the time series problems with instructor and group members. • Enter the Bitcoins’ data into FCAR model and writing result for an academic paper.