I built machine learning models that help optimize platforms like Google Ads and Facebook Ads to more efficiently spend digital media dollars. Provided business intelligence by means of statistical modeling and testing, to help Direct to Consumer brands better understand their customers purchase behaviors and make predictions/recommendations about best practices.

I construct models for binary and multi-level classification problems on various fields of university student data. Python packages such as numpy, scipy, pandas, and scikit-learn are used to preprocess, fit, and predict. Cross validation is used regularly to ensure optimal model parameters are used in fitting the final model.

My Master’s Thesis at CSULB, to which, the primary research objective is classifying first time freshman students at risk of not graduating in four years or less. I use XGboost for growing an ensemble of shallow classification and regression trees, yielding a dynamic model which inherently minimizes the bias-variance tradeoff. Benefits of my research include identifying factors that attribute to graduation in a timely fashion, and producing a model used by higher education institutions in an effort to increase four year graduation rates.

I built a Classification Ensemble by Random Partitioning (CERP) package to predict a binary class level for high-dimensional feature spaces. R packages "rpart" and "ctree" are used in growing decision trees, and I wrote all the functionality necessary (i.e. Random Partitioning, Pruning, Majority Voting, and Model Evaluation Metrics) for the package. When used for classifying Leukemia Cancer, CERP out performed Random Forrest, ADABoost, and SVM yielding an Accuracy of 98.6.

I created this project to help people gather more complete NFL player data by making use of the NFL GAME API. I used Python’s Pandas package to equip data into a data frame and clean the data to fit a wide variety of data requests. Data requests were delivered as CSV files so the data could be accessible across multiple platforms (e.g. Excel, SAS, SPSS, etc.).

I worked on a team with the goal of forecasting a “total funding raised” value in real time for the Kickstarter crowd funded project Pebble Time. I was responsible for data collection/cleaning, fitting the Gradient Decent model, and cross validation, all in Python. Collecting data in real time, allowed for us to predict future “funds raised”, revealing insight on whether or not the project will reach its funding goals.

Efficiently and quickly solve common bioinformatics problems given by the popular Biotech webpage Rosalind. Problems include complementing DNA strands, counting nucleotides, finding GC content and implementing Mendel's First Law. Requires high level of programming skills as search/sort algorithms are used consistently.

Responsible for accurate pours, multi-tasking, and ensuring guest satisfaction as my number one priority.