A scientist, a mathematician, a programmer and an artist. Together, these four traits make me a problem solver and a creator — I get a thrill from crafting algorithms, and I love solving difficult problems. I graduated with degrees in physics and mathematics; two subjects that I am passionate about, and that introduced me to the power of data and statistical analysis.
I create custom algorithms to solve unique problems, and I use statistical analysis and machine learning to either mine data for useful insights, or create models that predict outcomes and/or reveal relationships between attributes of a subject. The possible applications are nearly limitless, and specifics vary with each project.
The bottom line: I optimize the performance of businesses and systems, make predictions & generate insights, and enable new methods with custom algorithms — all using scientific and mathematical methods.
The nutshell — I am capable of working on a wide variety of problems, with data sets that are either raw or clean, big or small. I recommend looking at some of my projects for specific examples.
A few business-relevant topics that fall within my abilities:
The jargon — Some terms relevant to my abilities are: regression, classification, optimization, cluster analysis, time series, ANOVA, data exploration, pre-processing, feature engineering, working with data imbalance, categorical with many levels, sparse matrices.
I am familiar with many machine learning algorithms, but my favorites are Artificial Neural Networks, Random Forest Classifiers and Naive Bayes.
I can develop just about anything using Python and its libraries — including (but not limited to) Theano, Scikit-Learn, NumPy, Pandas, Requests and Matplotlib. I am fully capable of setting up and querying SQL databases, and I have a great deal of experience with web-oriented languages.
A few of my coding projects can be found below, along with some of my data projects.
Not sure what can I do for you?
Let's start a dialogue! Submit 3 fields so I can contact you with initial thoughts, and we can go from there (Next business day reponse). Regular contact here.
The links below lead to pages with the contents of semi-formal reports for some of my projects. In the case of the coding projects, each page contains a summary of the program with an outline of some features and a link to the code on GitHub.
An automated method of producing sales forecasts for approximately three thousand individual departments, using 75 ARIMA models.
Credit Card Fraud Detection & Cost MinimizationDetecting fraudulent transactions and using Bayesian minimum risk with calibrated probabilities for decision making.
Naive Bayes for Breast Cancer DiagnosticsTesting a Gaussian Naive Bayes Classifier by diagnosing cell masses as benign or malignant, and evaluating feature importances.
Marketing Campaign AnalysisInferring the effects of marketing channel spending and gender targeting on unit sales, using multiple regression analysis.
A full-featured program to construct, train, and evaluate convolutional or fully-connected neural networks.
Naive Bayes ClassifierA program for classification using quantitative and qualitative data using Naive Bayes.
Multiple Regression AlgorithmPerforms OLS multiple regression, ANOVA, F-tests, and t-tests.