Projects

My project repositories can be found at:

ZipLee

repo: github.com/liljonnystyle/sf_traffic

My personal Capstone Project in development at Zipfian Academy. Through careful feature engineering, cars are clustered into distinct driving behaviors, and network graphs are populated with customized edge weights. A webapp is deployed to recommend optimal driving routes for each cluster for any given trip in San Francisco. The project can be extended to incorporate live traffic feeds, predict fuel economy, and even distribute traffic loads.

Instagram Mosaic

repo: github.com/liljonnystyle/insta_mosaic

My first foray into web scraping and SQL databases. I wrote a python script with mechanize and BeautifulSoup to methodically crawl through Instagram and download images from a network of users. Ultimately, >13,200 images were collected in a few hours. A 10x10 sampling of coarse-grained pixel intensity tuples were stored into a mySQL database for each image. Those samplings were then methodically matched (based on the dot product of pixel intensities) with pieces of Van Gogh's Starry Night.

Target Image: Van Gogh's Starry Night

Mosaic Image:

Parameter Sweep Results:

2048 Robot (on hiatus)

A robot that learns to play the infamous 2048 game. I wrote a framework in python of an artificial neural network to learn how to play the game strategically by watching a human player's moves. The main challenge remaining is to have python and javascript communicate back and forth.

(Courtesy of gabrielecirulli)