Portfolio

Projects

Below is a list (in no particular order) of recent projects I have completed.  Each project includes a link to the project code, most often in the form of a Jupyter Notebook.

Click-Through Rate Data Challenge

      • Problem – Given click through data for various origin pages, analyze click-through rates
      • Shows data exploration, data cleaning (detection and removal of erroneous data), analysis and visualization of click through rate, feature engineering using NLTK
      • Tech – Python, pandas, numpy, seaborn, NLTK
      • Secret Gist Link

Internet Ads Challenge

      • Problem – Supervised classification given large feature space, determine if an image is an ad or not.
      • Shows data exploration, data preparation, dimensionality reduction, random forest modeling
      • Tech – Python, pandas, seaborn, sklearn, numpy
      • Secret Gist Link

Loan Portfolio Challenge

      • Problem – Given a data set of delinquent loans, analyze current delinquency rates and predict which loans will have their delinquency improve or worsen.
      • Shows transition matrix, data exploration, and feature engineering through visualization.
      • Tech – Python, pandas, numpy, seaborn
      • Secret Gist Link

RentHop – Kaggle Competition

      • Problem – Given a set of features for a rental listing, predict how much interest (low, medium, high) a rental listing will receive.
      • Shows data exploration, feature engineering, data cleaning, random forest model, gradient boosted trees model, Kaggle submission.
      • Tech – Python, pandas, numpy, sklearn, xgboost
      • Gist Link

CRUD Web App – Class Project

      • Problem – Class project.  Given specs of a charitable organization design and build an SQL database, then create a web application to create, read, update, and delete items from the database.
      • Shows SQL used to create and manipulate a database, HTML to build website pages, Python to create the web application.
      • Tech – Python, Flask, HTML, SQL, MySQL
      • GitHub Link

Cognitive Builder Faire Hackathon – Four Square API

      • Problem – using IBM’s DSX and the Four Square API build something.  My team of 3 developed a set of neighborhood analytics with the goal being to answer the question of. “What is the best use of this empty lot?”
      • Shows working with Four Square’s API, converting data into a useful format, data visualization, geographic data visualization.
      • Tech – Python, pandas, folium
      • GitHub Link

 

Tutorials

Machine Learning Hello Word – Tutorial using the Iris Data Set

Kaggle Titanic Competition – A guide to your first Kaggle submission