Species’ ranges arise from the interplay between environmental preferences, biotic and abiotic environmental conditions, and accessibility. Understanding of – and predictive models on – species distributions often build from the assumption that these factors apply homogenously within each species, but there is growing evidence for individual variation. Here, I use movement data to investigate individual-level decisions and compromises regarding the different costs and benefits influencing individuals’ geographic locations, and the species-level spatial patterns that emerge from these. I first developed a new method that uses tracking data to quantify individual specialisation in geographic space (site fidelity) or in environmental space (environmental specialisation). Applying it to two species of albatrosses, I found evidence of site fidelity but weak environmental specialisation. My results have implications for how limited research efforts are best-targeted: if animals are generalists, effort are best spent by understanding in depth individual patterns, i.e., better to track fewer individuals for long periods of time; whereas if animals tend to be specialists, efforts should be dedicated to tracking as many individuals as possible, even if for shorter periods. I then investigated individual migratory strategies and their drivers in nine North American bird species, using ringing/recovery data. I found latitudinal redistribution of individuals within the breeding and non-breeding ranges that generally did not follow textbook patterns (‘chain migration’ or ‘leapfrog migration’). Migratory individuals tend to trade off the benefits of migration (better tracking of climatic niche; better access to resources) and its costs (increasing with migratory distance). I found that birds are more likely to remain as residents in areas with warmer winter temperatures, higher summer resource surpluses and higher human population densities (presumably because of a buffering effect of urban areas). Overall, my results highlight the importance of considering individual variation to understanding the ecological processes underpinning species’ spatial patterns.

At a competitive level, football teams often have multiple matches per week. Thus, time can be a limited resource for match analysts tasked with analysing the performance of their team and its opponents. Increased availability of data in the field offers possibilities to automate processes to save time. This thesis presents a method to automatically detect pre-defined moments of interest in the game, and how they can be analysed to gain insights into the play style of football teams and how they create and concede goal-scoring opportunities in open play. An algorithm was developed to synchronise Wyscout event data with Signality tracking data of 240 matches from the Swedish Allsvenskan, which resulted in a mean absolute error of 280 ms. Comprehensible features were extracted from the combined data to detect eleven moments of interest in the absence of manually labelled data: crosses, passes to the golden zone, switches of play, central through balls, wing plays, keeping possession, long balls from the back, counterattacks, establish possession, counterpresses and fall back. The detection for fall back failed, but the remaining moments of interest were detected with a precision of 0.84. The automatic moment detection was made accessible through a web-based application, enabling analysts to focus on analysing aspects of the game rather than spending time searching for them. The detected moments were then analysed, demonstrating that by conducting a more extensive analysis, a data-driven playbook providing insights into how football teams play and create or concede goal-scoring opportunities can be established. / På elitnivå har fotbollslag ofta flera matcher per vecka. Tiden kan därmed vara en begränsad resurs för matchanalytiker som har i uppgift att analysera prestationen hos sitt eget lag och dess motståndare. Ökad tillgänglighet av data i fotboll ger möjligheter att automatisera processer för att spara tid. Denna studie presenterar en metod för att automatiskt upptäcka fördefinierade spelfaser av intresse och hur de kan analyseras för att få insikter i fotbollslagens spelsätt och hur de skapar och släpper till målchanser i öppet spel. En algoritm utvecklades för att synkronisera Wyscout händelsedata med Signality spårningsdata från 240 matcher från Allsvenskan, vilket resulterade i ett genomsnittligt absolut synkroniseringsfel på 280 ms. Begripliga variabler extraherades från den kombinerade datan för att upptäcka elva spelfaser av intresse utan manuell datamärkning: inlägg, passningar till de gyllene zonerna, spelvändningar, centrala genomskärare, spel på kanterna, bibehållning av bollinnehav, långbollar från backlinjen, kontringar, etablering av bollinnehav, direkta återerövringar och tillbakafall. Detekteringen av tillbakafall misslyckades, men de återstående faserna av intresse upptäcktes med ett positivt prediktivt värde på 0,84. Den automatiserade fasupptäckten gjordes tillgänglig genom en webbaserad applikation, vilket möjliggör för analytiker att fokusera på att analysera olika aspekter av spelet i stället för att tillbringa tid på att söka efter dem. De upptäckta faserna av intresse analyserades sedan, vilket visade att genom att genomföra en mer omfattande analys kan en datadriven spelbok skapas som ger insikter i hur fotbollslag spelar och skapar eller släpper till målchanser.

Tracking algorithms provide the model to recognize objects’ motion in the past. Moreover, applied to an artificial intelligence algorithm, these algorithms allow, to some degree, the capacity to forecast the future position of an object. This thesis uses deep learning algorithms to predict the ball’s position in the three-dimensional (3D) Cartesian space given the players’ motion and referees on the 2D space. The algorithms implemented are the encoder-decoder attention-based Transformer and the Inception Time, which is comprised of an ensemble of Convolutional Neural Networks. They are compared to each other under different parametrizations to understand their ability to capture temporal and spatial aspects of the tracking data on the ball prediction. The Inception Time proved to be more inconsistent on different areas of the pitches, especially on the end-lines and corners, motivating the decision to choose the Transformer network as the optimal algorithm to predict the ball position since it achieved less volatile errors on the pitch. / Spårningsalgoritmer möjliggör för modellen att känna igen objekts tidigare rörelser. Dessutom om tillämpad till en Artificiell intelligensalgoritm, de tillåter till viss mån att prognostisera ett objekts framtida position. Detta examensarbete använder djupinlärningsalgoritmer för att förutsäga bollens position i det tredimensionella (3D) kartesiska utrymmet baserat på spelarnas och domarnas rörelse i 2D-rymden. De implementerade algoritmerna är den kodare-avkodare-uppmärksamhetsbaserade Transformer och Inception Time, som består av en sammansättning faltningsnätverk (CNN). De jämförs med varandra med olika parametriseringar för att se deras förmåga att fånga upp tidsmässiga och rumsliga aspekter av spårningsdata för att förutsäga bollens rörelse. Inception Time visade sig vara mer inkonsekvent på olika områden på planen. Det var extra tydligt på slutlinjerna och i hörnen. Det motiverade beslutet att välja Transformer-nätverket som den optimala algoritmen för att förutsäga bollpositionen, eftersom den resulterade i färre ojämna fel på planen.

Ce travail propose une contribution originale à la compréhension du comportement spatial des pêcheurs, basée sur les paradigmes de l'écologie comportementale et de l'écologie du mouvement. En s'appuyant sur des données du 'Vessel Monitoring System', nous étudions le comportement des pêcheurs d'anchois du Pérou à des échelles différentes: (1) les modes comportementaux au sein des voyages de pêche (i.e. recherche, pêche et trajet), (2) les patrons comportementaux parmi les voyages de pêche, (3) les patrons comportementaux par saison de pêche conditionnés par des scénarios écosystémiques et (4) les patrons spatiaux des positions de modes comportementaux, que nous utilisons pour la création de cartes de probabilité de présence d'anchois. Pour la première échelle, nous comparons plusieurs modèles Markoviens (modèles de Markov et semi-Markov cachés) et discriminatifs (forêts aléatoires, machines à vecteurs de support et réseaux de neurones artificiels) pour inférer les modes comportementaux associés aux trajectoires VMS. L'utilisation d'un ensemble de données pour lesquelles les modes comportementaux sont connus (grâce aux données collectées par des observateurs embarqués), nous permet d'entraîner les modèles dans un cadre supervisé et de les valider. Les modèles de semi-Markov cachés sont les plus performants, et sont retenus pour inférer les modes comportementaux sur l'ensemble de données VMS. Pour la deuxième échelle, nous caractérisons chaque voyage de pêche par plusieurs descripteurs, y compris le temps passé dans chaque mode comportemental. En utilisant une analyse de classification hiérarchique, les patrons des voyages de pêche sont classés en groupes associés à des zones de gestion, aux segments de la flottille et aux personnalités des capitaines. Pour la troisième échelle, nous analysons comment les conditions écologiques donnent forme au comportement des pêcheurs à l'échelle d'une saison de pêche. Via des analyses de co-inertie, nous trouvons des associations significatives entre les dynamiques spatiales des pêcheurs, des anchois et de l'environnement, et nous caractérisons la réponse comportementale des pêcheurs selon des scénarios environnementaux contrastés. Pour la quatrième échelle, nous étudions si le comportement spatial des pêcheurs reflète dans une certaine mesure la répartition spatiale de l'anchois. Nous construisons un indicateur de la présence d'anchois à l'aide des modes comportementaux géo-référencés inférés à partir des données VMS. Ce travail propose enfin une vision plus large du comportement de pêcheurs: les pêcheurs ne sont pas seulement des agents économiques, ils sont aussi des fourrageurs, conditionnés par la variabilité dans l'écosystème. Pour conclure, nous discutons de la façon dont ces résultats peuvent avoir de l'importance pour la gestion de la pêche, des analyses de comportement collectif et des modèles end-to-end. / This work proposes an original contribution to the understanding of fishermen spatial behavior, based on the behavioral ecology and movement ecology paradigms. Through the analysis of Vessel Monitoring System (VMS) data, we characterized the spatial behavior of Peruvian anchovy fishermen at different scales: (1) the behavioral modes within fishing trips (i.e., searching, fishing and cruising); (2) the behavioral patterns among fishing trips; (3) the behavioral patterns by fishing season conditioned by ecosystem scenarios; and (4) the computation of maps of anchovy presence proxy from the spatial patterns of behavioral mode positions. At the first scale considered, we compared several Markovian (hidden Markov and semi-Markov models) and discriminative models (random forests, support vector machines and artificial neural networks) for inferring the behavioral modes associated with VMS tracks. The models were trained under a supervised setting and validated using tracks for which behavioral modes were known (from on-board observers records). Hidden semi-Markov models performed better, and were retained for inferring the behavioral modes on the entire VMS dataset. At the second scale considered, each fishing trip was characterized by several features, including the time spent within each behavioral mode. Using a clustering analysis, fishing trip patterns were classified into groups associated to management zones, fleet segments and skippers' personalities. At the third scale considered, we analyzed how ecological conditions shaped fishermen behavior. By means of co-inertia analyses, we found significant associations between fishermen, anchovy and environmental spatial dynamics, and fishermen behavioral responses were characterized according to contrasted environmental scenarios. At the fourth scale considered, we investigated whether the spatial behavior of fishermen reflected to some extent the spatial distribution of anchovy. Finally, this work provides a wider view of fishermen behavior: fishermen are not only economic agents, but they are also foragers, constrained by ecosystem variability. To conclude, we discuss how these findings may be of importance for fisheries management, collective behavior analyses and end-to-end models.

In the popular sport of football, the exploration of key performance indicators has garnered significant interest among researchers, coaches, and analysts. While machine learning approaches, such as the expected goals model, have provided valuable insights into the attacking aspects of the game, the defensive side has received comparatively less attention. This thesis focuses on the defensive aspect of football, particularly during the opposition’s build-up phase, a strategy increasingly adopted by many teams. The goal of this project is to integrate valuable features from existing research with newly generated ones, developed in consultation with football experts, to create a model that provides insights into a team’s defensive behavior during the opponent’s build-up phase. The study utilizes synchronized event and tracking data from the Allsvenskan 2022 and 2023 seasons. An algorithm is developed to filter and analyze build-up sequences by generating appropriate defensive features. Subsequently, a logistic regression-based machine learning model is implemented to predict the outcome of an event during a build-up sequence, as well as the overall outcome of the sequence. This approach, enables the introduction of two new metrics aimed at evaluating a team’s defensive behavior during the opponent’s build-up phase. Additionally, a web-based application is developed to visualize and communicate the project results and insights to football experts and data analysts. Finally, the findings of this thesis highlight the benefits of combining tracking data with event data in football analytics. / Nyckeltalsundersökningar för att utvärdera och utveckla fotbollsklubbars prestation har väckt stort intresse bland forskare, tränare och analytiker. Traditionellt sett har dessa nyckeltal härletts genom observationsanalys. Dock har den ökande förekomsten av teknik inom den professionella fotbollsvärlden skapat möjligheter för att implementera mer automatiserade metoder för taktisk analys. Denna studie kommer specifikt att fördjupa sig inom det taktiska området av fotboll, som framstår som den mest relevanta och dynamiska delen av spelet. Målet med studien är att applicera modern teknologi, så som maskininlärning, på befintlig forskning, för att utveckla en modell som ger insikter om försvarsspelet under motståndarens uppbyggnadsfas. Studien undersöker synkroniserad händelse- och spårningsdata från Allsvenskan, Sveriges Högsta fotbollsserie, säsongerna 2022 och 2023. Efter att ha utvecklat en algoritm för att filtrera uppbyggnadssekvenser, användes denna data för att generera omfattande egenskaper som beskriver kvaliteten på ett lags försvarsspel. Därefter implementeras en maskininlärningsmodell, med hjälp av logistisk regressionsanalys, för att förutse utfallet av både en uppbyggnadssekvens och försvarsspelet. Resultatet visar på värdet av att kombinera spårningsdata med händelsedata inom fotbollsanalys. Modellens prestanda förbättrades avsevärt, både när det gäller att förutsäga utfallet av ett defensivt spel och en uppbyggnadssekvens.Dessutom har resultatet av studien lett till användbara insikter om försvarsspel för dataanalytiker inom fotboll. En webbaserad applikation utvecklades också för att visualisera och kommunicera resultaten.

This master thesis presents a machine learning framework for real-time gesture and skeleton-based action recognition, integrated with the Unity game engine. The system aims to enhance human-computer interaction (HCI) in gaming and 3D related applications through natural movement recognition, by training a model on skeleton tracking data. The framework is trained to accurately categorize and identify gestures such as kicks and punches, enabling a more immersive gaming experience not existing in traditional controllers. After studying the evolution of HCI and how machine learning has transformed and reshaped the interaction paradigm, the prototype system is built through data collection, augmenting, and preprocessing, followed by training and evaluating a Long Short-Term Memory (LSTM) neural network model for gesture classification. The model is integrated into Unity via Unity Sentis using Open Neural Network Exchange (ONNX) format, enabling efficient real-time action recognition in 3D space. Each component of the pipeline is available and adaptable for future custom- ization and needs, skeleton tracking and Unity integration is built using the ZED 2i camera and ZED SDK. Experimental results demonstrate that the system presented can achieve over 90% accuracy in identifying predefined gestures. As a bridging solution tailored for Unity, this framework offers a practical solution to action recognition that could be found useful in future applications. This work contributes to advancing human-computer interaction and offers a foundation for further development in gesture-based Unity game design.

This thesis focuses on self-tracking, which mean monitoring and recording information about oneself using digital technologies and its use by runners in the Czech Republic. The main aim of this thesis is to describe the Czech running community through a detailed sociological analysis, and to answer a question: Why runners use self-tracking and how they benefit from it? The theoretical part firstly deals with the topic of self-tracking itself, then examines the uses and gratifications theory and the theory of online communities. The analytical part is devoted to description, analysis and interpretation of the results of the author's own survey, in which 844 runners of whom 754 practice self-tracking participated. Several hypotheses concerning sociodemographic structure, running characteristics, motivation, gratifications and safety were formulated. Five key motives (self-control, orientation to result, self-improvement, habit and social interaction) that lead runners to use self-tracking devices were identified through exploratory factor analysis. The motives vary based on gender and running characteristics (experience with running, runner's level, frequency of running, trainer) that also represent the main influencing factor for self-tracking in general. Self-tracking is closely related to sharing...

This work shows the application of artificial intelligence (AI) on invasion game tracking data to realize a fast (sub-second) and adaptable search engine for sports scenes, scene ratings based on machine learning (ML) and computer-generated solutions using reinforcement learning (RL). We provide research results for all three areas. Benefits are expected for accelerated video analysis at professional sports clubs. / Diese Arbeit zeigt die Anwendung von künstlicher Intelligenz (KI) auf Invasionsspielverfolgungsdaten, um eine schnelle (unter einer Sekunde) und anpassungsfähige Suchmaschine für Sportszenen zu realisieren, Szenenbewertungen auf der Grundlage von maschinellem Lernen (ML) und computergenerierte Lösungen unter Verwendung von Verstärkungslernen (RL). Wir stellen Forschungsergebnisse für alle drei Bereiche vor. Es werden Vorteile für eine beschleunigte Videoanalyse in Profisportvereinen erwartet.