World Series Game 7 Twitter Trends

Photo credit: Matthew Roth

Social media sites commonly make data available to developers. By giving developers the ability to utilize APIs to access information, applications can be made that positively influence not only the developer, but the users of social media. By utilizing the Twitter API, I was able to stream tweets from Twitter (using Python) during Game 7 of the World Series between the Giants and the Royals. I saved the data into an excel file (.csv) which allowed me to analyze the data in R. The goal of this project was to view tweets per minute as they pertain to either the Giants or the Royals. If anyone wants this code just ask.

Giants vs Royals response to World Series game 7 as game progresses.

The start of the game corresponds to time = 0. At around 30 minutes into the game the Giants scored 2 runs. The Royals responded 5 minutes later scoring a run and again 5 minutes after that. At about the 1 hour mark the first World Series Challenge was completed and only the Giants fans cared. At around the 100 minute mark the Giants scored again and Madison Bumgarner entered the game. Royal’s fans attempted to will their team to victory. When Alex Gordon hit his triple at the end of the game Twitter exploded.

A Late Fantasy Football Preseason Prediction

photo credit Marianne O’Leary

Fantasy sports have taken America by storm and as a result has grown to become a multi-million dollar industry. The challenge for fantasy gamers is predicting players that will yield year long success. The issue for armchair GMs is that a player’s success is often dependent upon the play of their teammates (a wide receiver is dependent upon the quarterback looking his way). This article will focus on fantasy football quarterbacks, runningbacks and receivers. I apologize for the lateness of this article, but I did not want my league mates to view my findings.

Let me first start by confessing I have only participated in fantasy football twice, 2012 and 2013. In 2012 I did not have a strategy and finished about midway up the table thanks to the exploits of Peyton Manning. In 2013 I decided to develop a strategy which was to own as many quality runningbacks as possible, and as a result I won my PPR league. I was the proud owner of Reggie Bush, Ryan Matthews, Knowshon Moreno, Doug Martin, and Andre Brown. It also helped that I had Nick Foles, Colin Kaepernick, DeSean Jackson, Josh Gordon, Julian Edleman, and Larry Fitzgerald. Looking back now this team appears to be incredibly stacked, however, this roster was chalk full of risk on draft night. Reggie Bush was on a new team, Ryan Matthews was constantly hurt, Knowshon Moreno was considered to be the backup to Montee Ball, no one knew Chip Kelly’s offense would be amazing, and Josh Gordon was/is currently suspended. The bottom line is that there is some element of luck and timing, which is what makes a prediction so difficult.

I would like to begin by stating that I cannot see the future and that this presentation is only one persons view. All data was collected from ESPN. No rookies were considered in this project to eliminate assumptions about the level of play from college to professional. I also only analyzed 50 runningbacks, 80 receivers (including tightends), and 24 quarterbacks.

Before I began this project I compared ESPN’s 2013 rankings with their 2014 rankings to see if there is any carryover in preseason rankings. It was not incredibly surprising that a players rank from the previous year was not always the same as the previous year. This would indicate that that the previous year’s data does not tell the full story. In fact, as the season begins and progresses you get a better idea of team philosophy and how a particular player is to be utilized. That being said, I only have last year’s data to make my predictions. All analysis was performed using the statistical program R.

To see my attributes and rankings…

Free Online Courses are a Great Tool

This is my first post to this blog and the goal is to pass on my passion for analyzing data. I am not a Ph.D. student or even a grad student for that matter (perhaps some day). I am however a person who is addicted to learning about new and exciting ideas. I have a B.S. in biology from the University of South Florida where I studied biomechanics, check the research here. I am currently a student at Eastern Washington University (EWU) where I have nearly completed a second B.S. degree in electrical engineering.

So how does data factor into any of this?

Well in my pursuit of all things fascinating I began to quite literally obsess about artificial intelligence. I thought to myself, “self, how can I get into this?” Luckily right about this time I was hired to work at a small start-up company called RTneuro. The CEO Sanjay Kumar an incredibly intelligent and likable man who really gave me the opportunity to discover my potential as a data scientist told me about www.coursera.org and the incredible amount of free courses offered. I quickly signed up for a Machine Learning course taught by Stanford professor Andrew Ng. This course was a boot camp, but it gave me the tools to proceed into the great unknown.

I ventured into the great unknown with a classmate. We undertook a project that sought to collect data from a 3-axis accelerometer and using machine learning techniques were able to create an Android Application that could determine the user’s activity (rest, walk, or run) as well as performing user identification from attributes. Our professor who is a juggernaut in the field of digital signal processing thought we were insane. But we completed this project (I will post about this later) solely from knowledge collected from a free online course. I’m now taking the data science track courses offered by John Hopkins University and hope to continue developing fun projects. Maybe something to help me win my fantasy league…again.

My advice: free online courses are an excellent resource that all people who consider themselves students should exploit in a positive way.