Browsing by Author "Ojala, Markus"

Filter results by typing the first few letters
Now showing 1 - 7 of 7
  • Applying Bayesian Bandits For Solving Optimal Budget Allocation In Social Media Marketing
    (2017-06-08) Ahonen, Niko-Petteri
     Perustieteiden korkeakoulu | Master's thesis
    Sequential budget allocation problems in uncertain setting are hard to solve. Allocating marketing budget is one example of such problem. These problems are always balancing between exploiting the current best option and exploring if the option that seems best at the moment truly is the best. In probability theory these problems are called Multi-armed bandit -problems. Improvements in Bayesian computation have allowed Bayesian solution called random probability matching to this problem to be suggested. The idea is based on older idea presented by Thompson (1933). This thesis shows how this approach can be used in calculating optimal budget allocation and implements a automatic tool making budget optimization decisions in Facebook marketing.
  • Bayesian Workflow for Winning Bid Estimation in Real-Time Bidding
    (2023-12-11) Arsendi, Mattia
     Perustieteiden korkeakoulu | Master's thesis
    The introduction of Real-Time Bidding in the online advertising markets has increased the competitiveness between advertisers, given the large number of bidders participating in each auction. Therefore, developing a profitable bidding strategy represents a crucial challenge to making the advertisers’ business sustainable. In this work, a strategy for supporting advertisers in predicting the best possible bid under the condition of extreme uncertainty represented by the auction is presented. Specifically, the concept of the Utility function is introduced and applied in the context of advertisers’ profit. Therefore, a Bayesian workflow is proposed and developed to estimate the market’s bid distribution, which is essential for computing the bid that maximizes the expected Utility function. The profit accumulated using the proposed approach is shown, demonstrating the positive impact that adopting the presented method might have.
  • Randomization algorithms for assessing the significance of data mining results
    (2011) Ojala, Markus
     Perustieteiden korkeakoulu | Doctoral dissertation (article-based)
    Data mining is an interdisciplinary research area that develops general methods for finding interesting and useful knowledge from large collections of data. This thesis addresses from the computational point of view the problem of assessing whether the obtained data mining results are merely random artefacts in the data or something more interesting. In randomization based significance testing, a result is compared with the results obtained on randomized data. The randomized data are assumed to share some basic properties with the original data. To apply the randomization approach, the first step is to define these properties. The next step is to develop algorithms that can produce such randomizations. Results on the real data that clearly differ from the results on the randomized data are not directly explained by the studied properties of the data. In this thesis, new randomization methods are developed for four specific data mining scenarios. First, randomizing matrix data while preserving the distributions of values in rows and columns is studied. Next, a general randomization approach is introduced for iterative data mining. Randomization in multi-relational databases is also considered. Finally, a simple permutation method is given for assessing whether dependencies between features are exploited in classification. The properties of the new randomization methods are analyzed theoretically. Extensive experiments are performed on real and artificial datasets. The randomization methods introduced in this thesis are useful in various data mining applications. The methods work well on different types of data, are easy to use, and provide meaningful information to further improve and understand the data mining results.
  • Randomization of real-valued matrices for assessing the significance of data mining results
    (2008) Ojala, Markus
     Helsinki University of Technology | Master's thesis
    Tiedonlouhinta on tapa analysoida suuria määriä tietoaineistoa oleellisen tiedon löytämiseksi. Monet tiedonlouhinnan menetelmät soveltuvat reaaliarvoisten matriisien tutkimiseen. Tällaisia matriiseja esiintyy luonnostaan useissa sovelluskohteissa kuten bioinformatiikassa. Tässä diplomityössä tutkitaan reaalimatriiseista saatujen tiedonlouhinnan tulosten merkitsevyyden testausta. Työssä käytetään satunnaistukseen perustuvaa merkitsevyystestausta. Tulosta pidetään merkitsevänä, jos on epätodennäköistä saada vastaava tulos satunnaistetulla aineistolla, jolla on joitain yhteisiä ominaisuuksia alkuperäisen aineiston kanssa. Työssä omaksutaan lähestymistapa, jossa matriisin rivien ja sarakkeiden keskiarvot ja varianssit säilytetään satunnaistuksessa. Täten tiedonlouhinnan tulos on kiinnostava, jos se ei selity pelkästään matriisin rivien ja sarakkeiden keskiarvoilla ja variansseilla. Tässä diplomityössä kehitetään kolme menetelmää tällaisten satunnaisten matriisien tuottamiseksi. Menetelmiä analysoidaan sekä teoreettisesti että kokeellisesti, ja niiden näytetään olevan tehokkaita käytännössä. Menetelmien toimintakykyä arvioidaan sekä todellisella että keinotekoisella aineistolla. Työn tulokset näyttävät, että kehitetyt menetelmät ovat käyttökelpoisia tiedonlouhinnan tulosten merkitsevyyden määrittämisessä.
  • Refugee debate and networked framing in the hybrid media environment
    (2021-02) Pöyhtäri, Reeta; Nelimarkka, Matti; Nikunen, Kaarina; Ojala, Markus; Pantti, Mervi; Pääkkönen, Juho
     A1 Alkuperäisartikkeli tieteellisessä aikakauslehdessä
    In this article, we analyse how the debate on the ‘refugee crisis’ has been constructed in Finnish news media and social media by using big data analytics. The study applies big data with the aim of exploring the dynamics between the mainstream news media and social media and the ways in which these dynamics shape and strategically amplify different understandings of the refugee crisis. The research highlights over-emphasis of crime and threat-oriented themes on refugee issues in social media, as well as illuminates the distinct role of social media platforms in shaping debates through user practices of hyperlink sharing and networked framing. Together these findings suggest that the hybrid media environment provides a possible arena for polarization of the refugee debate that could also be used for political ends.
  • Searching for evolvable and nonevolvable functional relationships between genes
    (2010) Pettersson, Ville
     School of Science | Master's thesis
    The well known theory of evolution by Darwin explains how different species arise and evolve through the process of natural selection. Although natural selection can produce relatively complicated structures from simpler ancestors, there are limits to which structures can evolve and which cannot. In the article "Evolvability" Leslie G. Valiant studies the limits of natural selection from a mathematical point of view. He defines the concept of evolvability of functions, and shows that Boolean conjunction (AND) and disjunction (OR) functions are evolvable by the definition given in the article, whereas Boolean parity functions (XOR) are not evolvable. In this Master's Thesis we test if Valiant's theories are applicable to biological organisms. Namely, we study whether there is evidence for the existence of Boolean conjunction and disjunction functions and the nonexistence of Boolean parity functions in real life gene-expression data. The results show that functional relationships in gene-expression data roughly follow the evolvability results of Valiant's article, i.e., statistically significant conjunction and disjunction functions are abundant and parity functions are more rare. We conclude that Valiant's definition of evolvability can be a useful tool in studying the evolutionary properties of real life organisms and worth further research.
  • Tietueiden automaattinen tunnistus ja ryhmittely HTML-dokumenteista
    (2010) Palmunen, Juhana
     Informaatio- ja luonnontieteiden tiedekunta | Bachelor's thesis