Browsing by Author "Viitasaari, Lauri"

Filter results by typing the first few letters
Now showing 1 - 20 of 58
  • Applying SHAP to explain packet discards in LTE network
    (2020) Sood, Nitesh
     School of Business | Master's thesis
  • Assessing the covariates of property and violent crime in Helsinki with spatial controls
    (2018-08-24) Kaivola, Mikko
     Perustieteiden korkeakoulu | Bachelor's thesis
  • Deep Learning Object Detection Models in Robotic Process Automation
    (2020) Vuorimaa, Waltteri
     School of Business | Master's thesis
    In recent years, businesses have become increasingly aware of their need to automate previously manual business processes. Adoption rates of robotic process automation (RPA) tools have increased a lot, and this has generated a great deal of value for companies. But, alas, especially some older legacy applications are still extremely problematic from the point of view of RPA platforms, and they cannot often be interfaced in a same fashion as some more modern applications. This is especially lamentable, as it is often the older applications, which tend to be used in the more manual business processes. This thesis aims to build a foundation for building and evaluating commercializable computer vision (CV) object detection models, that resolve the issues faced by RPA applications that arise from the lack of operating system (OS) level user interface (UI) visual element data. This kind of models could be used as a basis for building completely new RPA tools, or for augmenting existing solutions. The research questions discussed by this study are as follows: 1. Is using machine learning models for the detection of user interface elements viable? 2. What are the implications to existing RPA applications? 3. Can we build a model that is able to generalize from data from another domain? This thesis discusses the theoretical background of object detection and machine learning, and presents three potential families of machine learning model architectures that could be used for the aforementioned tasks. These models were then trained and compared to the CV capabilities of a commercial RPA platform, UiPath. The results show that the most simple object detection models cannot compete with the current commercial offering. There is, however, a clear path to further research, and it is clear that building a object detection system for UI elements is viable. In addition, this thesis presents the first publicly available dataset of (manually) annotated desktop UI screenshot images, and a tool for generating artificial user interface data.
  • European Options and Local Times
    (2010) Viitasaari, Lauri
     School of Science | Master's thesis
    Modern mathematical finance is based on the methods of stochastic analysis and usually models apply martingale theory and stochastic integration theory. 111LSpecially, the well-known Black and Scholes model is based on Ito integrals and geometric Brownian motion. This study establishes the connection between Black and Scholes model and local time of geometric Brownian motion. First we introduce the Black and Scholes model and derive the basic properties of the market model. Another objective is to find a new integral representation for local time of geometric Brownian motion through the Black and Scholes model. We also study some applications. The results of the study are promising. Besides the new integral representation for local time, we derive two ways to compute the expectation of the local time of geometric Brownian motion. We applicate the integral representation to Black and Scholes differential equation and study the local time of exponential martingale. We also derive a formula for the price of European options which are determined by the difference of two convex functions. As an application of this, we show that in order to hedge a convex European option, the capital needed increases as the maturity of an option increases.
  • Evaluating machine-learning algorithms and strategies in sports betting context
    (2024) Korkee, Santeri
     School of Business | Master's thesis
    This thesis studies various different machine learning algorithms, and their perfor- mance in a sports betting context. In addition to this, some strategies for choosing the bets to take are evaluated. As a context, player propositions, or player props are the market studied. This is done because of the lack of studies, increased popularity of said market, and potentially higher edge on the bettor’s side. The testing is starts with getting the data and preparing it for the algorithms. This includes filtering, creation of new variables and feature selection. Three regression- based algorithms, along with three classification algorithms are chosen. Of them, multiple iterations are created and tested with different datasets. The main objec- tive is to find a system that would consistently beat the sportsbooks by gaining prof- its in a simulation that is run in a realistic setting, thus showing the inefficiency of the market. In addition to that, the best bet picking strategy and algorithmic tech- niques are looked into. The objective set was reached: using correct strategy, multiple iterations were able to make noteworthy profits in the simulation, using 2023-24 season as the test. Not only were they very profitable, but also stable, which reduces the risk of bankruptcy. Other findings included the superior performance of regression-based algorithms compared to the classifiers, and the essentiality of finding a good strategy for pick- ing the bets. For the strategy, one based on maximal difference from the sports- books’ prediction was used.
  • Evolution and Impact of Executive Compensation: Historical Insights and Modern Implications
    (2024) Hietanen, Tuukka
     School of Business | Bachelor's thesis
    Executive compensation is a rapidly evolving topic that has garnered significant attention in recent years. It has raised major questions regarding the fairness of executive pay and its cost to shareholders. This thesis examines the history of executive compensation after the Great Depression through a comprehensive literature review. The research highlights key elements that have led executive compensation to rise to its modern level. The literature review also focuses on the current situation and examines the different forms of executive pay and analyzes the issues surrounding executive remuneration. The findings suggest that executive compensation has risen significantly, especially after the 1980s, largely due to the increase in stock-related compensation, mainly in the form of stock options. Furthermore, the findings illustrate the discontent that people have regarding executive compensation. Moreover, the efforts that governments have made to combat managerial remuneration have shown to be problematic if not even counterproductive at times.
  • Examining the link between footwear fit features and customer preference in footwear comfort: case study of running shoes on Letsrun online platform
    (2020) Nguyen, Thu
     School of Business | Master's thesis
    Abstract Purpose: Mass customization helps adapt to customers’ preference and maximize customer satisfaction. However, there are undoubtedly several challenges towards realizing this concept, especially in the footwear industry. Based on the case study of a footwear feedback platform, this Master’s thesis identifies the key footwear fit features that have an impact on customer preference in comfort and gives recommendation in terms of footwear design in running shoes. Design: An ordinal regression model is built based on the online customer reviews of running shoes on Letsrun platform to analyze the relationship between comfort rating and the variations in the footwear fit features. Findings: Narrower-than-medium heel width and narrower-than-medium forefoot width are found to have negative correlation with footwear comfort preference, while lower-than-medium toe box height and wider-than-medium heel width have a significant positive effect on footwear comfort preference. In addition, personal arch height is found to moderate the relationships between shoes last shapes, forefoot width, toe box height and the footwear comfort preference expressed by the runners. Research limitations: While certain footwear fit features are found to influence runners’ comfort preference substantially, the conceptual model is still limited in the explanatory power when we check the model robustness through the OLS method. More design features can be included to improve on the model. The generalization possibility and the number of observations are still limited. Practical implication: Shoe designers and manufacturers should pay attention to the heel width, forefoot width and toe box height features to improve customer comfort in general. Moreover, customizing running shoes based on arch height is essential to improve runners’ comfort, specifically for high-arched runners. Originality/Value: This study is the first empirical study to examine the relationship between footwear fit design features and customer preference in comfort level. The study complements literature in product design and mass customization in footwear products. Keywords Mass customization; Running shoes; Customer preference; Comfort; Product design
  • Flexible integrated functional depths
    (2021-02) Nagy, Stanislav; Helander, Sami; van Bever, Germain; Viitasaari, Lauri; Ilmonen, Pauliina
     A1 Alkuperäisartikkeli tieteellisessä aikakauslehdessä
    This paper develops a new class of functional depths. A generic member of this class is coined Jth order kth moment integrated depth. It is based on the distribution of the cross-sectional halfspace depth of a function in the marginal evaluations (in time) of the random process. Asymptotic properties of the proposed depths are provided: we show that they are uniformly consistent and satisfy an inequality related to the law of the iterated logarithm. Moreover, limiting distributions are derived under mild regularity assumptions. The versatility displayed by the new class of depths makes them particularly amenable for capturing important features of functional distributions. This is illustrated in supervised learning, where we show that the corresponding maximum depth classifiers outperform classical competitors.
  • Flexible transition probability model for assessing cost-effectiveness of breast cancer screening extension to include women aged 45-49 and 70-74
    (2023-06-23) Shafik, Nourhan; Ilmonen, Pauliina; Viitasaari, Lauri; Sarkeala, Tytti; Heinävaara, Sirpa
     A1 Alkuperäisartikkeli tieteellisessä aikakauslehdessä
    Breast cancer is the most common cancer among Western women. Fortunately, organized screening has reduced breast cancer mortality. New recommendation by the European Union suggests extending screening with mammography from 50-69-year-old women to 45-74-year-old women. However, before extending screening to new age groups, it's essential to carefully consider the benefits and costs locally as circumstances vary between different regions and/or countries. We propose a new approach to assess cost-effectiveness of breast cancer screening for a long-ongoing program with incomplete historical screening data. The new model is called flexible stage distribution model. It is based on estimating the breast cancer incidence and stage distributions of breast cancer cases under different screening strategies. The model parameters, for each considered age group, include incidence rates under screening/non-screening, probability distribution among different stages, survival by stages, and treatment costs. Out of these parameters, we use the available data to estimate survival rates and treatment costs, while the modelling is done for incidence rates and stage distributions under screening policies for which the data is not available. In the model, an ongoing screening strategy may be used as a baseline and other screening strategies may be incorporated by changes in the incidence rates. The model is flexible, as it enables to apply different approaches for estimating the altered stage distributions. We apply the proposed flexible stage distribution model for assessing incremental cost of extending the current biennial breast cancer screening to younger and older target ages in Finland.
  • Forecasting electricity prices in Finland: A comparative study of ARIMA and GRNN models
    (2024) Ouyang, Chengcheng
     School of Business | Master's thesis
    In the deregulated electricity market, uncertainty about electricity price trends is increasing. Especially in recent years, electricity prices fluctuate frequently and dramatically with accompanying price crisis in Finland. Therefore, market participants are exposed to higher risks. By leveraging appropriate electricity price forecasting models, they could make informed decisions to mitigate risks. In this study, ARIMA models and GRNN models are applied to forecast electricity prices in Finland. The objective is to explore how the electricity price forecasting models perform. Moreover, there are comparisons between ARIMA models and GRNN models to discuss the optimal models in different situations. The findings demonstrate that both models perform better in normal price period than in crisis. After comparing these two types of models, it is found that whether in normal or crisis situations, ARIMA models yield more accurate forecasts than GRNN model.
  • From generalized linear models to logistic regression: a theoretical overview
    (2019-05-22) Vienola, Nuutti
     Perustieteiden korkeakoulu | Bachelor's thesis
  • Gene regulatory network inference from sparsely sampled noisy data
    (2020-07-13) Aalto, Atte; Viitasaari, Lauri; Ilmonen, Pauliina; Mombaerts, Laurent; Gonçalves, Jorge
     A1 Alkuperäisartikkeli tieteellisessä aikakauslehdessä
    The complexity of biological systems is encoded in gene regulatory networks. Unravelling this intricate web is a fundamental step in understanding the mechanisms of life and eventually developing efficient therapies to treat and cure diseases. The major obstacle in inferring gene regulatory networks is the lack of data. While time series data are nowadays widely available, they are typically noisy, with low sampling frequency and overall small number of samples. This paper develops a method called BINGO to specifically deal with these issues. Benchmarked with both real and simulated time-series data covering many different gene regulatory networks, BINGO clearly and consistently outperforms state-of-the-art methods. The novelty of BINGO lies in a nonparametric approach featuring statistical sampling of continuous gene expression profiles. BINGO’s superior performance and ease of use, even by non-specialists, make gene regulatory network inference available to any researcher, helping to decipher the complex mechanisms of life.
  • Generative AI in the Workplace: Effects on Employee Efficiency and Performance
    (2024) Keltanen, Remu
     School of Business | Bachelor's thesis
    The objective of this thesis is to explore how the use of generative artificial intelligence (AI) in the workplace affects employee performance. Through a comprehensive review of existing literature and research on the topic, the findings suggest that generative AI has the potential to enhance employee productivity across various tasks by automating routine processes, freeing up time for more complex activities, improving output quality, reducing skill gaps among employees, and fostering creativity and innovation. However, the use of generative AI in the workplace also presents challenges, including the need to understand its capabilities, recognize false information, address potential negative impacts on high-skilled workers, and navigate privacy, security, and the efficient utilization of this technology.
  • The impact of a busy road on apartment values - A network-level approach
    (2020) Rujala, Henri
     School of Business | Master's thesis
    In this thesis, two-stage least squares regression with Gaussian distance decay kernel weight was employed to investigate the connection between road type and apartment values in Helsinki, Finland. This work contributes to the past domestic literature by including demographic variables as explanatory variables and measuring the road-type specific impact that noise has on apartment values. The research questions were the following: (1) There exists a statistically significant relationship between road type and apartment values. (2) Apartments located on the busy road have a discount when compared to apartments located on the quiet roads. (3) Apartments located near motorways or trunk roads have a discount when compared to apartments that are not located near motorways or trunk roads. The results of this thesis are fully confirming the research questions of (1) and (2). The research question of (3) was partially confirmed. According to the results, there exists a statistically significant relationship between road-type and apartment values. Also, when the apartment is located within a busy road, there is a discount of 1.9 percent present. Moreover, when the apartment is located within 250 meters from trunk road or motorway, there is a discount of 4.1 percent present. On the other hand, when the apartment is located within 500 meters from the trunk road or motorway, there is an increase of 1.9 percent in apartment values.
  • The Impact of Point-Based Incentive Programs in Decentralized Applications: A Case Study of YOLO Games Casino on the Blast Blockchain
    (2024) Messo, Vilppu
     School of Business | Bachelor's thesis
    This thesis investigates the impact of a point-based token incentive program on the usage of decentralized applications (dApps). It focuses on YOLO Games, a decentralized casino on the Blast blockchain. The study uses a case study approach to analyze the impact of incentive programs implemented by YOLO Games and Blast on the betting volume of YOLO Games. Interrupted Time Series (ITS) analysis assesses the key events influencing bet volume and their statistical significance. The findings suggest that changes in incentive structures, such as fee adjustments and changes in point systems, significantly affect bet volume. This research contributes to understanding the dynamics of incentive-driven user behavior in blockchain-based applications.
  • Integration in a Normal World: Fractional Brownian Motion and Beyond
    (2014) Viitasaari, Lauri
     School of Science | Doctoral dissertation (article-based)
    This thesis is about stochastic integration with respect to Gaussian processes that are notsemimartingales. Firstly, we study approximations of integrals with respect to fractionalBrownian motion and derive an upper bound for an average approximation error. Secondly, westudy the existence of pathwise integrals with respect to a wide class of Gaussian processes andintegrands. We prove the existence of two different notions of pathwise integrals. Moreover,these two different integrals coincide. As an application of these results, the thesis containsintegral representations for arbitrary random variables. Finally, we study a certain modelinvolving a Gaussian process and provide estimators for different parameters. We applyMalliavin calculus and divergence integrals to obtain central limit theorems for our estimators.
  • Latent model extreme value index estimation
    (2024-07) Virta, Joni; Lietzén, Niko; Viitasaari, Lauri; Ilmonen, Pauliina
     A1 Alkuperäisartikkeli tieteellisessä aikakauslehdessä
    We propose a novel strategy for multivariate extreme value index estimation. In applications such as finance, volatility and risk of multivariate time series are often driven by the same underlying factors. To estimate the latent risks, we apply a two-stage procedure. First, a set of independent latent series is estimated using a method of latent variable analysis. Then, univariate risk measures are estimated individually for the latent series. We provide conditions under which the effect of the latent model estimation to the asymptotic behavior of the risk estimators is negligible. Simulations illustrate the theory under both i.i.d. and dependent data, and an application into currency exchange rate data shows that the method is able to discover extreme behavior not found by component-wise analysis of the original series.
  • Lean Management in Surgical Practices
    (2024) Anttila, Joonatan
     School of Business | Bachelor's thesis
    Lean is a set of tools originally developed to enhance efficiency of industrial manufacturing. Since the 2000s, Lean has been increasingly implemented into healthcare practices. Surgery is a major cost and resource utilization contributors in healthcare system. This narrative literature review examines the application of Lean methodology within surgical context by reviewing relevant case studies. For the relevant literature Scopus and PubMed databases were searched to identify case studies implementing lean directly into perioperative system. Seven case studies, spanning various fields of surgery, were selected for inclusion in this review. The results show that Lean has been successfully implemented to surgical practices evidenced by the case studies. Most studies reported improved key parameters, including reduced turnover times, decreased number of complications, shortened perioperative times, enhanced team morale and cost reductions. As a conclusion Lean management methodologies show considerable potential to improve the efficiency of surgical practices. Challenges within implementation are associated to possible resistance to change and need of extensive staff training. Future high-quality studies eliminating systemic bias are needed to further examine possibilities of Lean in surgical context.
  • Local times and sample path properties of the Rosenblatt process
    (2021-01) Kerchev, George; Nourdin, Ivan; Saksman, Eero; Viitasaari, Lauri
     A1 Alkuperäisartikkeli tieteellisessä aikakauslehdessä
    Let Z=(Zt)t≥0 be the Rosenblatt process with Hurst index H∈(1∕2,1). We prove joint continuity for the local time of Z, and establish Hölder conditions for the local time. These results are then used to study the irregularity of the sample paths of Z. Based on analogy with similar known results in the case of fractional Brownian motion, we believe our results are sharp. A main ingredient of our proof is a rather delicate spectral analysis of arbitrary linear combinations of integral operators, which arise from the representation of the Rosenblatt process as an element in the second chaos.
  • Methods for Removing Digital Image Noise by Merging Multiple Exposures
    (2019-08-30) Uusihärkälä, Viljami
     Perustieteiden korkeakoulu | Bachelor's thesis