Using PageRank to Rank Boxers
https://www.insidescience.org/news/using-math-rank-boxers
As discussed in lecture, Google’s PageRank algorithm is best known for its ability to retrieve the most relevant webpage for what’s being searched. I understood why PageRank was so important, especially in terms search engine optimization, but I wondered what other real world applications PageRank could assist with. That’s when I discovered Yuen Yiu’s article which investigates the use of PageRank algorithms in ranking with World’s greatest heavyweight boxers.
In more detail, PageRank determines a webpage’s influence based on the number of links it receives from other webpages. A link from one page to another is considered a vote, and pages with a higher number of votes are considered more important, thus the votes they cast to other pages carry more weight.
In terms of sport rankings, most team sports have extended playoff periods in which top teams that were successful during the regular season face off against one another until the top two face off in a championship match. This easily determines a team’s rank. Boxing, on the other hand, is more complex. Top boxers aren’t guaranteed an opportunity to go toe-to-toe. Different sanctioning bodies impose different regulations on the sport, which causes a disconnect between the amount of points each boxer is awarded per fight. For some bouts, the amount of points the winner is awarded is dependent on how many points their opponent possesses. Similarly, some boxers receive more points for a knockout than others, and some participants lose more points for being inactive than others. Due to the lack of matches and continuity within the sport, boxing is a very difficult sport to effectively rank. That was, until Tien Chih and Demitri Plessas applied Google’s PageRank method to boxing ranks. Looking at heavyweight boxers from 1990-1999, Chih and Plessas linked all boxers through their individual wins and losses. Just as Google places different valuations for pages that are referred to more often, and by whom their being referred by, so too did Plessas and Chih. They looked at how often a boxer was victorious, how victorious their opponent was and so on. By using wins as the links/referrals between each boxer, they found out “who’s the man who beats the man who beats the man…” and thus were able to rank the top 20 heavyweight fighters. Through PageRank they determined that Evander Holyfield ranked first among heavyweight boxers between 1990 and 1999.
This article discusses just one of PageRanks many applicable uses. In class we talked about PageRank in more web-based conceptual terms, which made it a little difficult for me to grasp at first. But now, after looking through many articles discussing the real world application of the PageRank algorithm-something that can help rank just about any object, organization or industry-the concept of PageRank became much more clear. Due to it’s accuracy, and easy implementation, PageRank is being taught as a helpful tool within data science.