Imagine that you have spent your entire life working towards one thing; a dream that you have had since childhood. It is your life and your identity. With hard work and perseverance, you have achieved it, but you have put all your eggs in one basket. Not only are you relying on yourself for future financial stability, but your entire city and team is counting on you as well. This is the life of most NHL hockey players. NHL Success determines a player's future in life and a team's future in their cities and in the overall sport. Many players retire before their mid-thirties, with few having a college degree. This means that many rely on their savings from their time in the league to maintain their lifestyle after the NHL. Thus, success in hockey not only determines their current earnings but also determines the savings that they will rely on for the rest of their lives. Additionally, teams need to be successful to not only stay in their city but also to stay in the sport, exemplified by the Atlanta Thrashers becoming the Winnipeg Jets due to financial problems. Without success and a strong fanbase supporting the team, franchises will struggle financially and be unable to continue operating in their current state. This leaves the question what constitutes success in hockey and what steps can the teams undertake to understand current trends in the public and the league that can help guide future decisions.
Hockey player's success is quantified by salary, which determines not only their lifestyle but also their future. But, what makes hockey players successful? Many factors can be analyzed to help predict success. Is it physical size, is it country of origin, is it birth date, is it goals, or location of team which pays the best? Many believe that bigger players are stronger and better for a physical sport like hockey. Other followers of hockey contend that smaller players tend to be nimbler and faster on the ice. Which is better? Certain countries such as Canada are known for their hockey players and being the founders of the sport, but do certain European countries produce a small number of elites? Some argue that the higher number of goals directly determines success, but does being a team player and having large number of assists matter more? These hypotheses guided the analysis and provided a background to make conclusions on achieving success in hockey.
After the unsupervised learning method of clustering, initial conclusions were made on the factors of height, goals, assists and earnings. The analysis found that the taller a person was the more money they earned. Additionally, it was the combination of high number of goals and assists that determined the players who were paid the most. This suggests that bigger players are payed more because they have the ability to take have higher impact in a high contact sport such as hockey. This results in more chances of interfering with the other team or helping one's team get possession, due to their physical size, negatively impacted the other team. All of these factors ultimately make the bigger player valuable to their team. Furthermore, the results showed that although goals are important, being a team player and setting others up with assists resulted in more earnings as seen below.
This could be because the more a player passes and sets others up, the more they control the game and provide greater opportunities for more goals during the game when compared to one just scoring and not passing. This provides an explanation of why it is a combination instead of pure goals. These conclusions are supported by other models.
These supervised models were able to differentiate between those who had an uneven assists and goals ratio compared to those who had an even assist to goal ratio, then placing them in either the low or high earning groups. This supports the idea that those who have uneven ratio are low earners while those with even ratios are high earners. Adding to the analysis, the Decision Trees indicated that although teams have similar salaries, those who play in larger cities are more likely to be paid more. This is likely due to the larger city teams being more competitive, attracting better players, and having better finances due to higher ticket revenue compared to smaller city teams. Furthermore, Canadians dominate the league with more players but with a wider range of player salary. Whereas, players from European countries are more likely to be in the higher paid range of salaries. Europeans have a more polarized salary standard, either they are paid the highest amount or earn the lowest. Canadians have the highest base pay but have more average players then standout stars
From this data, it can be concluded that only the best of the best and the ones with clear talent from other countries are able to play in the league, while Canadians have a higher average skill which is used to fill roster numbers. This idea is further supported by the supervised learning that was able to successfully group players based off of their biographic information.
Franchises need the support of their fan base. Thus, analyzing and differentiating twitter data concerning the NHL and news about the NHL is imperative. Can trends in twitter data about certain decisions, games, or players be found in these tweets? Although twitter data can be random and seemingly unconnected, certain thoughts, ideas, and feelings can be inferred from analysis on them.
Initially twitter data using unsupervised models were unable to differentiate between the overarching topics of the tweets determined by the three hashtags NHL, NHLNews and NHLFanatasy. These results may be because the tweets are very similar and only differ slightly, thus, leading to the lack of distinct clusters based off of the topics. However, in the supervised models, the models were able to differentiate the difference because the labels allowed them to determine the slight differences between them. Furthermore, certain topics and issues that were currently occurring in the NHL were able to be determined. A specific example of this is in #NHL the tweets analysis indicated how the public reacted to the first home game for the New York Rangers. It showed that fans were extremely excited for the first game, which indicates that the tweets can indicate public enthusiasm or energy toward certain events, thus, guiding teams in their decision making. Additionally, words such as injury, miss and percentage determined from the analysis can help teams determine public thoughts on the injuries and how players are performing, providing them with additional information to help determine player contracts. Finally, the analysis was able to differentiate the NHL tweets into two categories such as NHLnews and NHL hashtags. This model can be used to separate tweets centered around the league into these two categories to determine the thoughts of the public on the league in general or on specific news such as player contracts, performance or management decisions. This can guide a franchise's decisions in the future based on public opinion consequently protecting their interest and the fans interests, while maximizing the benefits for both sides.
Overall stats and tweets can guide decisions for both players and franchises. As the league gets more competitive as skill level increases players and teams need to find the edge over their competition from their stats to public perception. Hockey success is a dream to many but it is something that few are able to achieve and maintain.