38-1. It’s a record that any team would covet at this point in the college basketball season. Unfortunately, that one small blemish in an otherwise pristine record came at the wrong time for the Kentucky Wildcats. Perhaps equally distressing, this loss not only ended an historic, perfect season prematurely, but it will also linger in the collective memory of many Americans. According to Turner, last night’s broadcast by Turner Sports and CSB Sports garnered the highest cable rating ever for a college basketball game, and it was the highest rated Final Four game since 1993.
For those of you who missed it, the result to last night’s memorable Final Four matchup between Kentucky and Wisconsin in the Men’s NCAA Basketball Tournament came down to the wire, with the game tied at 60-60 with 2:25 left in the match. Flashback to the 6:36 mark in the second half, and it would’ve been easy to call the game. Kentucky’s large frontcourt and speedy guards had overcome a 52-44 deficit to build a 60-56 lead against the Badgers, and their second half resurgence meant that they had a 79.2% chance of advancing to the championship game tomorrow. According to Nate Silver’s simulations, if both teams had twelve remaining possessions at a 60-56 scoreline, Kentucky’s winning percentage was as high as 81.9%.
Instead, Wisconsin’s defense held Kentucky scoreless between the 6:36 and 0:56 stretch. Now, AP Player of the year, Frank Kaminsky, has an opportunity to completely redeem last year’s Final Four loss by playing an in-form Duke team, which handily defeated Michigan State earlier in the day, in tomorrow’s championship game.
Luckily, this writer had the opportunity to animate a map of Tweets correlating to this intense game. This process consisted of three steps:
- data mining with the R statistical computing language using the filterStream function;
- exporting, cleaning, and filtering the dataset on Excel; and
- using the “Torque” and “Intensity” features of CartoDb, a cloud platform for GIS and web mapping tools, to interactively map that dataset.
Here are some interesting facts and observations pertaining to the maps and their creation:
- Each tweet had to include at least one of the eleven following keywords: “Wisconsin,” “WisconsinBadgers,” “Badgers,” “Kentucky,” “KentuckyWildcats,” “Wildcats,” “WisconsinvsKentucky,” “KentuckyvsWisconsin,” “UKvsWIS,” “Wisconsin vs Kentucky,” “FinalFour,” and “NCAATournament.”
- The TwitteR filterStream search mined a total of 303,403 worldwide tweets from Twitter (130,228 first half tweets and 173,175 second half tweets) during the game. The filterStream function was used throughout the game, but not during halftime. This would explain the blank gap in the map about halfway through the video.
- Of the 300K+ Tweets that were randomly mined, 12,224 were location enabled tweets. In other words, the Twitter user had location enabled for Tweets, so those specific Tweets could be mapped with given Longitude and Latitude specifications and creation times. Therefore, the random plots on these maps represent approximately 4% of the data collected.
- The game was played at a relatively controlled manner. At halftime, both teams were deadlocked at 36 points apiece, and neither team managed to achieve a double-digit lead at any point in the game. Unsurprisingly, the steady generation of tweets in the video matches what was largely a steady, calculated game. In addition, it is important to note that unlike certain sports, such as football or soccer, where scoring often occurs in isolated spurts, basketball features scored baskets throughout a game. Barring a seminal event like a game-winning shot, one would not expect a drastic surge of Tweets at any point during the game.