NBA Prediction Model:
An Overhead Look at My Deep Learning Ensemble Methodology
Dec. 22nd 2017
Shallow Neural Networks
Dec. 16th 2017
Neural Networks Basics
Dec. 11th 2017
Building an Application:
Live NBA Scoreboard
Dec. 7th 2017
One Month Later in Cleveland:
Defense, Offense, Clutch Shots, and Respect Where Respect is Due
Right about now I'm pretty happy I concluded my last Cavaliers article with the opinion that things would change in Cleveland because since I strongly critiqued LeBron's lack of effort they have gone 13-0, fueled by LBJ domination and clutch shot after clutch shot. Let's take a look at how things have turned around over the past month.
Dec. 9th 2017
Scraping NBA Line Movements with BeautifulSoup4
Here I wanted to take a step backward and walk through how I created the database that was used in my post exploring Plotly's Dash platform by creating an dashbaord that displays realtime NBA line movements. The application runs constantly in the background my linux server using cascading crontabs and was written completely in Python using BeautifulSoup4 for the web scraping and Postgres queries to store the data.
Dec. 5th 2017
NBA Team Rankings:
As of last night 30 NBA teams have officially played 1/4 of their 2017-2018 Season and it seems fitting to look back at this first quarter and give my input on where each team lies. This is my first attempt at a "Power Rankings" article and I plan on writing one at the conclusion of each quarter. Rather than simply giving you a subjective ranking spot I go through a few salient metrics for each team and try to back up some of my at least partially biased logic. For each team I also give a "Worth it to Watch (WITW)" ranking because having a high winning percentage doesn't always correlate to being interesting to watch. If you're familiar with Zach Lowe's "League Pass Rankings" you get the picture.
Dec. 3rd 2017
Introduction to Deep Learning
It has been almost a year since I finished Andrew Ng's famous course on Machine Learning. It lived up to it's immense hype and then some, so naturally when I started to get the itch to take a new class I quickly stumbled across his deep learning, 5 course AI specilization track and figured I'd give it a try. Over the next few weeks, in and around my research, blogging, and job hunt I plan to log my progress through the course here in a series of posts.
Nov. 25th 2017
Dash Framework Exploration:
Live NBA Line Movements
Always in search of new data viszualization methods and tools, I recently came across Dash, a Python framework built on top of Plotly.js, React, and Flask with the purpose of building analytical web applications completely in Python. Dash works as the frontend to our analytical Python backend. In order to get a feel for the platform I decided to build out a sample application that uses many of Dash's features.
Nov. 14th 2017
The Atrocious Defense of the Cleveland Cavaliers
And How LeBron James Isn't Helping
It's getting late here on the east coast and I just finished watching the defending Eastern Conference Champions rack up their 7th loss of the season and 10th straight game giving up at least 110 points, ratcheting up their Defensive Efficiency to an absurdly porous 113.0, good for the worst in the NBA and on pace for one of the worst of all time. Of course we are only 1/7th into the season, and per-usual the defensive sinkholes that are JR Smith and Kevin Love continue to show zero effort on the defensive end, but there is a bigger and much more important reason why the Cavs are such a joke on defense: LeBron James. In this post I want to walk through the early-season disaster that is Cleveland's defensive effort and see if this is a bit more worrying than the standard "not trying before April" excuse the Cavs always seem to get.
Nov. 9th 2017
E1Q1: KNN Classifiers
This morning I want to test out my method for integrating ipython notebook work into blog posts while simultaniously brushing a little rust off my understanding of Convolutional Neural Nets. Lets do this by working through the first exercise in Stanford's CS231n: Convolutional Neural Networks for Visual Recognition class. Much of the material is familiar to me but a lot of the application is new as I have not worked extensivly on applying ML to imaging problems so it seems like a good way to get the gears turning!
Aug. 22nd 2017