Deadlifts and Derivatives

Updates and Research from Steve Bronder

The purpose of this site is to give current information on Steve Bronder's (me) research and personal life

Thanks for coming to my page! My name is Steve Bronder and the purpose of this website is to present updates in my research and personal life. Currently, I am doing Data Science stuff at Capital One. Previously, I attended Columbia University for a Master's in Quantitative Methods. I was born and raised in Pittsburgh, Pennsylvania and graduated from Duquesne University with a major in Economics and a minor in Quantitative Analysis. My previous experience falls in the category of R, C++, and python package development, machine learning, data visualization, forecasting, and predictive analytics.  My main hobbies include programming, dancing, four square, comic books and Magic: The Gathering.


Contact Information

Current Location: 

New York City

Email: 

sbronder @ this website


 Below are the list of languages I have used with small descriptions on what I use them for and whether I prefer it.

Technical
Proficiencies

R - This is my most proficient language with about 5 years of experience.  I wrote the package PANICr in my senior year of college which is available on CRAN, have given talks on predictive modeling in R, and know my way around building interactive graphs. I also have a forecasting extension to mlr that automates a large part of the model development process.

C/C++ -  I use both for for CUDA and OpenCL research. C is of course, C. It's fine, but I'm only using it if I really need to. I generally get things to compile and do what I want in C++ and sometimes I would even be so bold as to say I write modern C++.

Fortran - Column major order makes much more sense! I'm currently on a 'Make Fortran Great Again' kick.  For the class Computational Statistics I wrote a collapsed Gibbs sampler in Fortran. It's a very sleek and powerful language if your only goal is to make math go fast.

VBA - A link in my research section hosts a genetic algorithms written in VBA that lets you play tic-tac-toe against a robot. It's fine, but not my preference.

LaTeX - I love pretty graphs and type case therefore I love LaTeX. There is no easier way to impress someone than with good typography.

SAS - I learned SAS during my time at Management Science Associates.  Performing quality assurance checks at MSA taught me PROC SQL, macros, DATA steps, etc. SAS is important for statistical analysis, but I would rather use R when possible.

 

 

 

Python -  At Capital One I've used python to build a package for the collection, processing, and validation of data from very manual sources. My groups work focuses on Collateralized Loan Obligation style deals and the intake data is very very messy.

For the class Computational Statistics at Columbia I designed a module for Latent Dirichlet Allocation in python, with a Fortran back engine to do the heavy lifting. In my GPU computing course we use Python as a wrapper language to easily implement CUDA code.

SQL - I use SQL in my day to day work. It's fine

Linux -   I run ubuntu on both my laptop and desktop and know my way around a terminal. Writing bash scripts, automating jobs with pcron, and whatnot

Microsoft Office - Word, Excel, Powerpoint, Access. I know them.

Maple - I've used Maple for constrained optimization in my paper "An Analysis of the Effects of Machinery in Labor Markets". Maple is effective and I feel comfortable in its environment.

Matlab - I used Matlab for my paper "Aggregation Bias in the US Personal Consumption Expenditure." I am comfortable with Matlab. A large chunk of my R package was written by translating Matlab code into R.

 


Work 

 

Predictive Analyst Intern; Zurich Insurance                                      May 2016 - August 2016

At Zurich Insurance, my work focused on model evaluation and the development of performance metrics. Using my model validation strategies, I was able to build a model to predict losses on automotive insurance contracts that beat industry standards. My research at Zurich insurance taught me that good model evaluation comes from a rigorous and reproducible methodology as well as creating loss functions that are specific to your projects needs.

Marketing Analyst; Management Science Associates                   May 2014 - August 2014  and May 2015 - August 2015

One of my first duties at MSA was using my knowledge of VBA to build data visualization tools for the client. The client was able to select several inputs at once that would then be placed into a graph with summary information printed out for each variable.  Once I became comfortable with the client's product and market I was tasked with performing quality assurance on data sets. I used SAS's PROC SQL, custom macros, and DATA steps to perform these QA's.

As an R user, MSA gave me the opportunity to work in the research and development section of the business analytics department in order to work on Bayesian Hierarchical models and data visualization using R. I was able to move the research timeline forward by successfully implementing a two layer Bayesian Hierarchical model for a demand elasticity study before the end of the summer. I also wrote a tutorial on graphing, scraping coordinate data, and mapping coordinate data in R and presented an hour long 'brown bag' where I showed everyone how to create interactive graphics in R with Rcharts and D3. Tutorials for these can both be found in my Research section.

Research Assistant; Economics Department                              January 2014 - December 2014

A local company has given Duquesne's Econ department access to their customer data. Our task is to develop an elasticity model to point out categories in decline and categories that are undervalued. The data set is over 1TB which gave me the opportunity to use R in a big data context. I have been able to work around computational constraints by using out-of-memory algorithms provided by Revolution R Enterprise. The end goal of this project is to implement an algorithm to analyze demand elasticity of products in real time. My part in the project included building a script to automate the cleaning and merging of new data sets as they became available.


Activities

President and Founder                                                      August 2011- August 2013

     Duquesne Young Americans For Liberty 

In the beginning of my sophomore year I started a Young Americans for Liberty chapter on my campus in order to spur political and philosophic conversation on my, at the time, politically moot campus. Through organizing tabling events, meetings, and bringing in speakers our group was able to slowly turn Duquesne into a politically active campus, headed by the now yearly Duquesne Political Science Honors Society Debate. My co-founder and I coordinated with other groups at nearby colleges in order to bring current political issues into the minds of students throughout our respective campuses.