I'm a lifelong student who is passionate about data science and sustainability. I have over 6 years of experience performing data engineering, data analysis, and building machine learning models to provide business intelligence solutions. Additionally, I have been working in the commercial building energy and sustainability sector for over 11 years and I am passionate about developing resources that contribute to more energy efficient designs.
I developed, tested, and currently maintain the pyExpandObjects repository, which acts as a pre-processor program that maps simplified JSON template objects into complex components for building energy modeling simulations (EnergyPlus).
I am in the process of developing tools to extract information from the U.S. Energy Information Administration (EIA) bulk download feature, and make some initial predictions to gain insight.
ETL Details - PySpark
ETL Output Analysis
Coal Net Generation Analysis
Electricity Net Generation Data Exploration
I, like many others, have found much value in the book Forecasting: Principles and Practice. The code that comes with that text provides a very sophisticated cross-validation technique that is packaged as the tsCV() function. However, when used with external predictors (xreg), the function does not always return results. The detailed breakdown of the bugs, and proposed fix, are provided in this summary report. Additionally, a vectorized version of this function is provided which may work on larger data sets but is untested.
This is a comprehensive report that explores and transforms a set of survey questions about commercial building characteristics, extracts the most impactful features pertaining to electrical and natural gas consumption, and uses those features to create a deep feed-forward neural network prediction algorithm for each fuel source.
This project was created in an effort to showcase some programming and visualization skills I acquired. It is built using HTML, CSS, Javascript (d3.js), and the Plotly graphing package. It is a little slow, as it is hosted on a free platform, but it serves the purpose of showing what can be visualized from such data.