Me and My Humble Portfolio

My name is Alex.

I am a final year graduate student undertaking Business Analytics at Columbia University, currently pursuing a career in the data science field.

Over the past 4 years, I have been actively delivering data-driven solutions in both professional and research settings. Along the way, I have led my teams to win Best Insights Award at DataFest, a datathon sponsored by the American Statistical Association, and to achieve 3rd place in a Kaggle competition. This summer, I had a great time working alongside some brilliant minds at Activision Blizzard, looking into players' in-game behavior.

I use the newest data science techniques to turn vast quantities of data into valuable insights that help decision-makers maximize impact.

Check out my Resume here!

Topics I Have Covered

Since the embark on my adventure of data science, I have worked on projects on...

Regression Analysis
Classification Problems
Nature Language Processing
Computer Vision
Anomaly Detection
Cluster Analysis
Survival Analysis
Sports Analytics

Skills I Have Accquired

Through working on all those projects, I have learned to use...

if skills in ['Python', 
               'SQL', 
               'Scala', 
               'R', 
               'Tableau', 
               'C++',
               'Git', 
               'Apache Spark', 
               'Google Cloud']:

    proficient_knowledge = True

else:
    willing_to_learn = True

Projects I Have Done

It’s through our actions — our decisions — that we affect the world around us.

Churn Analysis： Customer-At-Risk Model for account managers

Delivered a Churn Model and made recommendations for user retention strategies.
Potential uplift in profit: $150K

Survival Analysis on Battle Royale Mode in Call of Duty: Black Ops4

Customized Exponential Accelerated Failure Time Model to examine players' survival prospect in game.
Used as a reference for designers to balance game

NLP: Generating Knowledge Graphs to Predict Fertilizer Price Trends

Parsed entities, and trained word embeddings using tools including DBpedia Spotlight annotation, Stanford CoreNLP, Word2vec, and etc. in parallel.
Automated insights from mineral reports using NLP

The Data Open: Improving Community Health with Renewable Energy Source

Applied Panel Regression, Breusch-Pagan test, the Hausman test, and the Granger Causality test to examine the relationship between non-renewable energy consumption and asthma prevalence.
Completed analysis and 12-page-long paper in a time limit of 8 hours

Recommendations on Usage of LinkedIn Sales Navigator

Scraped LinkedIn profiles to uncover insights on characteristics of converted clients with machine learning and topic modeling methods, and advised on pitching different potential clients using LinkedIn Sales Navigator
Increased conversion rate by 50%

Game Analytics: Identify Cheaters in Game PlayerUnknown’s Battlegrounds (PUBG)

Implemented One-Class SVM and Isolation Forest to detect cheaters, and defined metrics to evaluate the results of unsupervised learning models.
Identified cheaters with a true positive rate of 64.39%

Computer Vision: Video Analytics

Overlaid AI logic rules to boost detection accuracy and reduce false alarm rate, and created a pipeline to deliver scheduled analytical reports.
Increased detection accuracy by 7.6%, and reduced false alarm rate by 5.9%