Skip to content

stefanbund/MADSPortfolio

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

48 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Portfolio of Data Science Projects for Stefan Bund

During my degree at University of Michigan I produced these data science projects (Masters in Applied Data Science, MADS). I have also included research I did as principal at rad disco.

Predicting Ideal Trades in Volatile Markets

Predicting short term price movements in volatile cryptocurrency markets. Used a machine learning pipeline to deliver highly accurate (97%) trade predictions for year 2022-23 in the Avalanche crypto market (AVAX-USD) on Coinbase. Built to support real time Highly Frequent Trade environments.

Solving Workforce Attrition: Using Causal Inference for a Case Study Client

Presented a solution for solving workforce atttrition at a model, case study client. I used a logistic regression model to discover most likely causes of employee turnover at an Indian manufacturing firm. Utilized a correlation matrix to discover highly correlated principal compenents, which were fed into an ROC/AUC analysis.

Predictive Factors in Sell-Side Metaorders: Capitalization Flows as Precursor to Price Shifts

A study in asset prices, and tightly correlated behavior between sell volume and pessimistic economic announcements.

Can the United States Decouple from Chinese Exports?

A study on the American dependence on Chinese exports. Proposes margins by which the US Trade Office may negotiate new trade agreements with nations within the Comprehensive and Progressive Agreement for Trans Pacific Partnership member nations, per industry sector.

Locating Highly Covariant Commodity Trade Clusters

I used python to discover affine clusters of tradeable commodities, and investigate how a major firm such as Citadel will use affine clusters of financial products to predict pricing.

Directd Acyclic Graphs for Heartrate Efficiency and Holistic Fitness

I investigated how the implementation of various fitness devices accompanied improved workout efficiency, as measured through workout intensity and heart rate graphs. Used during an introduction to data visualization course with Chris Brooks.

Grammar of Graphics for Time Series, and Twitter Streams

A means of embedding highly expressive ASCI graphics (emoji) into altair area charts. A method for locating the most-expressive tweeted emoji for a real-time sports event.

Custom Test Train Splits for Improved Random Forest Accuracy

A proprietary method for unbiased test train splits, improving Random Forest model accuracy. Used to reduce model complexity in unsupervised learning projects, where high degrees of dimensional complexity must be resolved, in addition to preserving unbiased samples among sequential data sets. Used to resolve a prediction model inside a personal fitness context.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published