Manager Data Analyst

Shuxi Lian

301-***-**** ● https://www.linkedin.com/in/shuxi-lian/ ● Greenbelt, Maryland 20770 ● *********@*****.*** EDUCATION

University of Maryland, Robert H. Smith School of Business College Park, MD Master of Science in Business Analytics, GPA 3.63/4.0 December 2019 Relevant Courses: Database Management, Data Mining, Big Data and AI, Pricing and Revenue Management Jimei University, Navigation College Xiamen, Fujian, China Bachelor of Engineering, Transportation (International Shipping Management) June 2018 Honors: University Scholarship in 8 semesters, 3rd Prize in National English Public Speaking Contest TECHNICAL SKILLS

• Programming languages: Python (numpy, pandas, scikit-learn, scipy, keras), R, SQL, VBA, JavaScript(D3)

• Tools: Excel, PowerPoint, Tableau, Google Analytics, AWS

• Statistical Modeling: Machine Learning, Neural Networks, Natural Language Processing WORK EXPERIENCE

Institute for Development Impact (Management Consulting Company) Washington D.C. Data Analyst Intern Jun 2019 – August 2019

• Researched classification criteria from WHO by leveraging Beautiful Soup and Regex in Python for 1000 survey data

• Prepared data dictionary of 350 statistics, stored results on GitHub, preprocessed missing and error data

• Ran 10 statistical tests in Chi-squared test and T-test, identify correlations and presented results to the manager

• Automated analysis that includes customized pivot table function, statistic visualization, water practice distribution, 10 statistical tests in Chi-squared test and T-test in Jupyter Notebook for research of next 5 years

• Collaborated with co-workers utilizing Git, created an online interactive visualization using JavaScript library D3, published Radial Bar Chart as an open source package in Node Package Manager Xiamen Hengxingsheng Trade Co., Ltd (International Trade Company) Xiamen, Fujian, China Business Analyst Intern May 2017 – August 2017

• Researched on-market raw material suppliers, leveraged MySQL to extract 2-year historical direct cost data to produce reports covering price fluctuation and revenue growth

• Utilized advanced Excel to analyze price of raw materials and products line; developed and automated interactive dashboard using Tableau for the sales manager to present during Trade Fair PROJECT EXPERIENCE

Regime Modeling with Text Feature for Financial Service Industry

• Extracted 20-year 308 text features and regime data from database using SQL; applied feature engineering based on domain knowledge, correlation, recursive feature elimination and LASSO to reduce the dimension of data to 39

• Utilized Logistic Regression, Random Forest, XGBoost and deep learning models to do two-class classification, predict 5-week lags regime labels with highest accuracy of 85%, evaluate the model based on AIC and p-value

• Used feature importance to identify 10 important features which contribute to the change of market regime Google Analytics - Experiential Learning with Real Client in E-commerce

• Worked in a team of 6 people and client to develop, implement, and execute marketing strategies to increase sales

• Designed marketing campaigns with Google Ads, created 118 ad groups with 15,898 keywords for different target product, tracked and analyzed customer behavior such as demographics and bounce rate with Google Analytics

• Acquired 20 new customers, engaged over 554 active users, and increased client’s revenue by 10% within 3 weeks Geospatial Data Analysis for Crimes

• Cleaned dataset and classified crime types into three categories based on Uniform Crime Report in Python

• Converted address into geocode using GeoPy, visualized it on Tableau map to present heatmap of risk

• Built a KNN model to predict the crime prone area with respect to time of day, day of week and month with an accuracy of 54%; visualized the prediction using a geographical heat map by color Deep Learning for Photo Aesthetics Analysis

• Used data augmentation to avoid overfitting by rotating and flipping images; normalized 8000 images to same size

• Led a team of 3 to apply convolutional neural network to train 3 models by adjusting parameters of hidden layers

• Applied ensemble method to eliminate bias of single model and achieved 42% accuracy DISTINCTIONS

• Fluent in Mandarin, English and Hokkien-Taiwanese ● Public speaking, British Parliament Debate

Contact this candidate