Browsing by Author "Korpimies, Samuel"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Predicting players' success on the PGA-Tour
    (2020) Korpimies, Samuel; Malo, Pekka; Viitasaari, Lauri; Tieto- ja palvelujohtamisen laitos; Kauppakorkeakoulu; School of Business
    The PGA-Tour is the most prestigious golf tournament circle in the world and being able to predict outcomes of the tournaments is continuously on the mind of bookmakers, bettors, and academics. This thesis attempts to use the official PGA-Tour statistics gathered from 2010 to 2019 to predict the success of the players in future tournaments. The literature review concentrates on the existing academic research about what statistics correlate most with success in professional golf and how to even measure success in professional golf. In this thesis a player’s success is measured by the number of Top 10 finishes he gets during the 2019-season. The original data from the PGA-Tour could not be used as such but needed to be heavily modified in order to fit the two models that were used to make the predictions. PCA-feature-clustering was applied to bundle some of the original variables to reduce the multicollinearity issues in the training data. The models used to predict the probabilities, with which players reach the top 10, are Logistic Regression and Random Forest Classifier. Assessing how well the models performed was done by calculating the average deviation of each player’s predicted percentage of Top 10 Finishes by the end of the 2019 season compared to the actual percentage of Top 10 Finishes. The Random Forest Classifier model performed better, than the older Logistic Regression model, with a mean AD of 4.486 percent. This means, that on average, the percentage of Top 10 Finishes predicted for each player was only off by 4.486 percent. In conclusion, using machine learning algorithms, the PGA-Tour statistics could be used to predict the future success of players. However, the accuracy of the models could be improved by tweaking them more or using newer and more complex machine learning algorithms.
  • Thumbnail Image
    Item
    Sponsorships in ESports
    (2017) Korpimies, Samuel; Fodness, Dale; Mikkelin kampus; Kauppakorkeakoulu; School of Business
    Objectives The main objectives of this study were to research how sponsorships in eSports are conducted. The research expands on the current knowledge that exists about how eSponsorships are acquired and how a successful sponsorship partnership is maintained. Additionally the ways of how to measure effectiveness of an eSports sponsorship are to be explored. Summary Three (3) eSports sponsors/sponsees were interviewed using semi-constructed interviews. The acquisition process of sponsorships in eSports varies from sponsor to sponsor depending on the company’s objectives. Sponsors have trouble measuring the success of their sponsorship and have to rely on very few techniques about how to measure the value created by the partnership. The main finding about how to maintain a working relationship between the sponsor and sponsee is to have open discussions if any trouble arises. Conclusions The research concluded with a more in-depth look into how processes in eSponsorships are constructed. The size of the sponsor and eSports team dedicates which party has more power in the relationship. Sponsors and sponsees have to be flexible when it comes to deliverables. Otherwise, minor problems can ruin the relationship between the partners.