Щоб переглянути інші типи публікацій з цієї теми, перейдіть за посиланням: Prediction Explanation.
Дисертації з теми "Prediction Explanation"
Оформте джерело за APA, MLA, Chicago, Harvard та іншими стилями
Ознайомтеся з топ-29 дисертацій для дослідження на тему "Prediction Explanation".
Біля кожної праці в переліку літератури доступна кнопка «Додати до бібліографії». Скористайтеся нею – і ми автоматично оформимо бібліографічне посилання на обрану працю в потрібному вам стилі цитування: APA, MLA, «Гарвард», «Чикаго», «Ванкувер» тощо.
Також ви можете завантажити повний текст наукової публікації у форматі «.pdf» та прочитати онлайн анотацію до роботи, якщо відповідні параметри наявні в метаданих.
Переглядайте дисертації для різних дисциплін та оформлюйте правильно вашу бібліографію.
1
Gordon, Richard Douglas. "Explanation and prediction in the labour process theory." Thesis, University of British Columbia, 1990. http://hdl.handle.net/2429/30583.
Повний текст джерелаАнотація:
The view that large-scale, long-range social theories cannot be predictive other than "in principle" is sufficiently widespread as to be considered the orthodox view. It is widely held that, lacking this predictive quality, social theories are cut off from a crucial form of vindication enjoyed by the experimental sciences. Thus many would agree with Ryan's assessment that while with regard to large-scale social changes "long-range prediction is not in principle impossible," nonetheless as a matter of practical methodology such a goal is of "dubious value."
The reason commonly proffered as to why social theories cannot be predictive is the causal complexity of social life. Because of this feature, it is held, while we may be able to unearth interesting social generalizations, we will not be able to predict the many initial conditions together with which they predict. Alternately, due to this complexity we are able to achieve no better than tendency laws which do not permit predictions of sufficient precision to allow for predictive testing. This has been held to be true for other causally complex fields as well. Thus, Scriven has argued that Darwin was "the paradigm of the explanatory but non-predictive scientist" due to the constraints imposed on his methodology by the causal complexity of the biosphere.
As a result of both an uncritical acceptance of the orthodox view and an
inadequate analysis of Marx's methodology, Daniel Little has argued that Marxian theory is non-predictive. However, a thorough analysis of Marx's labour process theory shows it to be both clearly predictive and subject to justification by predictive assessment. Moreover, a formalization of the theory indicates that available data confirm it as regards both its central hypothesis and the matrix of social causation it exhibits.
Little's position in regard to Marxian theory is strongly similar to Scriven's in regard to Darwinian theory. In both cases, faulty theoretical presuppositions combine with inadequate analysis to buttress false conclusions as to the asymmetry of explanation and prediction. Adequate analysis dispels Little's and Scriven's conclusions and exhibits important methodological parallels between Marx and Darwin.<br>Arts, Faculty of<br>Philosophy, Department of<br>Graduate
2
Bonawitz, Elizabeth R. (Elizabeth Robbin). "The rational child : theories and evidence in prediction, exploration, and explanation." Thesis, Massachusetts Institute of Technology, 2009. http://hdl.handle.net/1721.1/47891.
Повний текст джерелаАнотація:
Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Brain and Cognitive Sciences, 2009.<br>Includes bibliographical references (p. 122-133).<br>In this thesis, rational Bayesian models and the Theory-theory are bridged to explore ways in which children can be described as Bayesian scientists. I investigate what it means for children to take a rational approach to processes that support learning. In particular, I present empirical studies that show children making rational predictions, exploration, and explanations. I test the claim that differences in prior beliefs or changes in the observed evidence should affect these behaviors. The studies presented in this thesis encompass two manipulations: in some conditions, children's prior beliefs are equal, but the patterns of evidence are varied; in other conditions, children observe identical evidence but children's prior beliefs are varied. I incorporate an additional approach in this thesis, testing children within a variety of domains, tapping into their intuitive theories of biological kinds, psychosomatic illness, balance, and physical systems. Chapter One introduces the problem. Chapter Two explores how evidence and children's strong beliefs about biological events and psychosomatic illness influence their forced-choice explanations in a story-book task. Chapter Three presents a training study to further investigate the developmental differences discussed in Chapter Two. Chapter Four looks at how children's strong differential beliefs of balance interact with evidence to affect their predictions, play, explanations, and learning.<br>(cont.) Chapter Five looks at children's exploratory play with a jack-in-the-box, (where children don't have strong, differential beliefs), given different patterns of evidence. Chapter Six investigates children's explanations following theory-neutral evidence about a mechanical toy. Chapter Seven concludes the thesis. The following chapters will suggest that frameworks combining evidence and theories capture children's causal learning about the world.<br>by Elizabeth R. Bonawitz.<br>Ph.D.
3
Watson, Jason Paul 1971. "Explanation and prediction of curious experimental phenomena in lasers and nonlinear optics." Diss., The University of Arizona, 1999. http://hdl.handle.net/10150/282875.
Повний текст джерелаАнотація:
Experimental data often contains curious and unexplained results. In the course of experimental investigations of Raman shifting and the Co:MgF₂ laser, results were obtained which would not have been expected from the typical theoretical picture. In the case of Raman shifting, the forward Stokes conversion was found to depend upon the pump bandwidth. Numerical modeling suggests that coupling between the Stokes directions may be the root cause of the phenomena. In the case of the Co:MgF₂ laser, the laser output was observed to have large amounts of spectral structure. This amount of structure should not be expected in a room temperature vibronically broadened laser. Further experiments point to adsorbed water vapor for the cause of the structure, and this hypothesis is supported by a numerical model. Additionally, a unique method for treating the effects of arbitrary gain distribution on the propagation of the lowest order laser cavity mode is expanded to cover new distributions and new coordinate systems. An extension to parametric gains is also made. The extensions are then used to predict unstable regions in real laser cavities. These instabilities are observed in diffraction calculations. Guidelines for observing this intriguing result are presented.
4
Haar, D. H. "Formalised modelling of action theory in the explanation of crime for prediction, deduction and intervention." Thesis, University of Cambridge, 2009. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.599815.
Повний текст джерелаАнотація:
This dissertation proposes an original approach to theory of action in psychological and sociological criminology, i.e. to theory explaining the causation of human wilful behaviour at great abstraction through the information processing conducted by each individual human agent. It is argued that the model presented in this dissertation, the so-called Minimal Model of Action, is more theoretically comprehensive than prior familiar approaches originating in various related fields, in particular through its integration of both rational and habitual aspects of behaviour in a unified causal argument. Secondly, it is argued that the model is more methodologically appealing than previous approaches due to its formalisation through conventional mathematics. The proposed model is brought to bear on more concrete behavioural data and criminological problems in three separate chapters so as to scrutinise its validity and tractability from three methodologically different angles. An experimental chapter shows that empirical responses to computer-based scenario tasks frequently display behaviour patterns, especially forms of habituation, which the Minimal Model of Action in its simulated implementations and unlike previous models manages to explain and predict. In the following chapter, it is mainly shown through mathematical deduction both in continuation of and in juxtaposition to prior economic reasoning in which ways “optimum law enforcement” levels are systematically overestimated (and sometimes underestimated) under a variety of conditions when over-rationalised conceptions of the individual offender are employed. Finally, a chapter on aggregate levels of small-scale public corruption employs the general model to simulate a typical criminal phenomenon to the explanation of which economic and broader social conceptions of human agency equally should contribute.
5
McKay, William L. "Hope and suicide resilience in the prediction and explanation of suicidality experiences in university students." Laramie, Wyo. : University of Wyoming, 2007. http://proquest.umi.com/pqdweb?did=1456285751&sid=3&Fmt=2&clientId=18949&RQT=309&VName=PQD.
Повний текст джерела
6
Olofsson, Nina. "A Machine Learning Ensemble Approach to Churn Prediction : Developing and Comparing Local Explanation Models on Top of a Black-Box Classifier." Thesis, KTH, Skolan för datavetenskap och kommunikation (CSC), 2017. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-210565.
Повний текст джерелаАнотація:
Churn prediction methods are widely used in Customer Relationship Management and have proven to be valuable for retaining customers. To obtain a high predictive performance, recent studies rely on increasingly complex machine learning methods, such as ensemble or hybrid models. However, the more complex a model is, the more difficult it becomes to understand how decisions are actually made. Previous studies on machine learning interpretability have used a global perspective for understanding black-box models. This study explores the use of local explanation models for explaining the individual predictions of a Random Forest ensemble model. The churn prediction was studied on the users of Tink – a finance app. This thesis aims to take local explanations one step further by making comparisons between churn indicators of different user groups. Three sets of groups were created based on differences in three user features. The importance scores of all globally found churn indicators were then computed for each group with the help of local explanation models. The results showed that the groups did not have any significant differences regarding the globally most important churn indicators. Instead, differences were found for globally less important churn indicators, concerning the type of information that users stored in the app. In addition to comparing churn indicators between user groups, the result of this study was a well-performing Random Forest ensemble model with the ability of explaining the reason behind churn predictions for individual users. The model proved to be significantly better than a number of simpler models, with an average AUC of 0.93.<br>Metoder för att prediktera utträde är vanliga inom Customer Relationship Management och har visat sig vara värdefulla när det kommer till att behålla kunder. För att kunna prediktera utträde med så hög säkerhet som möjligt har den senasteforskningen fokuserat på alltmer komplexa maskininlärningsmodeller, såsom ensembler och hybridmodeller. En konsekvens av att ha alltmer komplexa modellerär dock att det blir svårare och svårare att förstå hur en viss modell har kommitfram till ett visst beslut. Tidigare studier inom maskininlärningsinterpretering har haft ett globalt perspektiv för att förklara svårförståeliga modeller. Denna studieutforskar lokala förklaringsmodeller för att förklara individuella beslut av en ensemblemodell känd som 'Random Forest'. Prediktionen av utträde studeras påanvändarna av Tink – en finansapp. Syftet med denna studie är att ta lokala förklaringsmodeller ett steg längre genomatt göra jämförelser av indikatorer för utträde mellan olika användargrupper. Totalt undersöktes tre par av grupper som påvisade skillnader i tre olika variabler. Sedan användes lokala förklaringsmodeller till att beräkna hur viktiga alla globaltfunna indikatorer för utträde var för respektive grupp. Resultaten visade att detinte fanns några signifikanta skillnader mellan grupperna gällande huvudindikatorerna för utträde. Istället visade resultaten skillnader i mindre viktiga indikatorer som hade att göra med den typ av information som lagras av användarna i appen. Förutom att undersöka skillnader i indikatorer för utträde resulterade dennastudie i en välfungerande modell för att prediktera utträde med förmågan attförklara individuella beslut. Random Forest-modellen visade sig vara signifikantbättre än ett antal enklare modeller, med ett AUC-värde på 0.93.
7
Hasan, Rakebul. "Prédire les performances des requêtes et expliquer les résultats pour assister la consommation de données liées." Thesis, Nice, 2014. http://www.theses.fr/2014NICE4082/document.
Повний текст джерелаАнотація:
Prédire les performances des requêtes et expliquer les résultats pour assister la consommation de données liées. Notre objectif est d'aider les utilisateurs à comprendre les performances d'interrogation SPARQL, les résultats de la requête, et dérivations sur les données liées. Pour aider les utilisateurs à comprendre les performances des requêtes, nous fournissons des prévisions de performances des requêtes sur la base de d’historique de requêtes et d'apprentissage symbolique. Nous n'utilisons pas de statistiques sur les données sous-jacentes à nos prévisions. Ce qui rend notre approche appropriée au Linked Data où les statistiques sont souvent absentes. Pour aider les utilisateurs des résultats de la requête dans leur compréhension, nous fournissons des explications de provenance. Nous présentons une approche sans annotation pour expliquer le “pourquoi” des résultats de la requête. Notre approche ne nécessite pas de reconception du processeur de requêtes, du modèle de données, ou du langage de requête. Nous utilisons SPARQL 1.1 pour générer la provenance en interrogeant les données, ce qui rend notre approche appropriée pour les données liées. Nous présentons également une étude sur les utilisateurs montrant l'impact des explications. Enfin, pour aider les utilisateurs à comprendre les dérivations sur les données liées, nous introduisons le concept d’explications liées. Nous publions les métadonnées d’explication comme des données liées. Cela permet d'expliquer les résultats en suivant les liens des données utilisées dans le calcul et les liens des explications. Nous présentons une extension de l'ontologie PROV W3C pour décrire les métadonnées d’explication. Nous présentons également une approche pour résumer ces explications et aider les utilisateurs à filtrer les explications<br>Our goal is to assist users in understanding SPARQL query performance, query results, and derivations on Linked Data. To help users in understanding query performance, we provide query performance predictions based on the query execution history. We present a machine learning approach to predict query performances. We do not use statistics about the underlying data for our predictions. This makes our approach suitable for the Linked Data scenario where statistics about the underlying data is often missing such as when the data is controlled by external parties. To help users in understanding query results, we provide provenance-based query result explanations. We present a non-annotation-based approach to generate why-provenance for SPARQL query results. Our approach does not require any re-engineering of the query processor, the data model, or the query language. We use the existing SPARQL 1.1 constructs to generate provenance by querying the data. This makes our approach suitable for Linked Data. We also present a user study to examine the impact of query result explanations. Finally to help users in understanding derivations on Linked Data, we introduce the concept of Linked Explanations. We publish explanation metadata as Linked Data. This allows explaining derived data in Linked Data by following the links of the data used in the derivation and the links of their explanation metadata. We present an extension of the W3C PROV ontology to describe explanation metadata. We also present an approach to summarize these explanations to help users filter information in the explanation, and have an understanding of what important information was used in the derivation
8
Rawstorne, Patrick. "A systematic analysis of the theory of reasoned action, the theory of planned behaviour and the technology acceptance model when applied to the prediction and explanation of information systems use in mandatory usage contexts." Access electronically, 2005. http://www.library.uow.edu.au/adt-NWU/public/adt-NWU20060815.154410/index.html.
Повний текст джерела
9
Bantegnie, Brice. "Eliminating propositional attitudes concepts." Thesis, Paris, Ecole normale supérieure, 2015. http://www.theses.fr/2015ENSU0020.
Повний текст джерелаАнотація:
Dans cette thèse je défends l'élimination des concepts d'attitudes propositionnelles. Dans le premier chapitre, je présente les thèses éliminativistes en philosophie de l'esprit et des sciences cognitives contemporaines. Il y a deux types d'éliminativisme: le matérialisme éliminatif et l'éliminativisme des concepts. Il est possible d'éliminer les concepts soit des théories naïves soit des théories scientifiques. L'éliminativisme à propos des concepts d'attitudes propositionnelles que je défends requière le second type d'élimination. Dans les trois chapitres suivants je donne trois arguments en faveur de cette thèse. Je commence par soutenir que la théorie interventionniste de la causalité ne fonde pas nos jugements de causalité mentale. Ensuite je montre que nos concepts d'attitudes propositionnelles ne sont pas des concepts d'espèces naturelles car ils groupent ensemble les états des différents modules d'une architecture massivement modulaire, la thèse de modularité massive faisant partie, je l'affirme, de notre meilleur programme de recherche. Finalement, mon troisième argument repose sur l’élimination du concept de contenu mental de nos théories. Dans les deux derniers chapitres de la thèse, je défends ce dernier argument. Tout d'abord, je réfute l'argument du succès selon lequel étant donné que les psychologues emploient le concept de contenu mental et ce faisant produisent de la bonne science ce concept ne devrait pas être éliminé. Ensuite je rejette une autre façon d'éliminer ce concept, celle choisie par les théoriciens de la cognition étendue. Pour cela je réfute le meilleur argument qui a été donné en faveur de cette thèse: l'argument du système<br>In this dissertation, I argue for the elimination of propositional attitudes concepts. In the first chapter I sketch the landscape of eliminativism in contemporary philosophy of mind and cognitive science. There are two kinds of eliminativism: eliminative materialism and concept eliminativism. One can further distinguish between folk and science eliminativism about concepts: whereas the former says that the concept should be eliminated from our folk theories, the latter says that the concept should be eliminated form our scientific theories. The eliminativism about propositional attitudes concepts I defend is a species of the latter. In the next three chapters I put forward three arguments for this thesis. I first argue that the interventionist theory of causation cannot lend credit to our claims of mental causation. I then support the thesis by showing that propositional attitudes concepts aren't natural kind concepts because they cross-cut the states of the modules posited by the thesis of massive modularity, a thesis which, I contend, is part of our best research-program. Finally, my third argument rests on science eliminativism about the concept of mental content. In the two last chapters of the dissertation I first defend the elimination of the concept of mental content from the success argument, according to which as psychologists produce successful science while using the concept of mental content, the concept should be conserved. Then, I dismiss an alternative way of eliminating the concept, that is, the way taken by proponents of extended cognition, by refuting what I take to be the best argument for extended cognition, namely, the system argument
10
Lonjarret, Corentin. "Sequential recommendation and explanations." Thesis, Lyon, 2021. http://theses.insa-lyon.fr/publication/2021LYSEI003/these.pdf.
Повний текст джерелаАнотація:
Ces dernière années, les systèmes de recommandation ont reçu beaucoup d'attention avec l'élaboration de nombreuses propositions qui tirent parti des nouvelles avancées dans les domaines du Machine Learning et du Deep Learning. Grâce à l'automatisation de la collecte des données des actions des utilisateurs tels que l'achat d'un objet, le visionnage d'un film ou le clic sur un article de presse, les systèmes de recommandation ont accès à de plus en plus d'information. Ces données sont des retours implicites des utilisateurs (appelé «~implicit feedback~» en anglais) et permettent de conserver l'ordre séquentiel des actions de l’utilisateur. C'est dans ce contexte qu'ont émergé les systèmes de recommandations qui prennent en compte l’aspect séquentiel des données. Le but de ces approches est de combiner les préférences des utilisateurs (le goût général de l’utilisateur) et la dynamique séquentielle (les tendances à court terme des actions de l'utilisateur) afin de prévoir la ou les prochaines actions d'un utilisateur. Dans cette thèse, nous étudions la recommandation séquentielle qui vise à prédire le prochain article/action de l'utilisateur à partir des retours implicites des utilisateurs. Notre principale contribution, REBUS, est un nouveau modèle dans lequel seuls les items sont projetés dans un espace euclidien d'une manière qui intègre et unifie les préférences de l'utilisateur et la dynamique séquentielle. Pour saisir la dynamique séquentielle, REBUS utilise des séquences fréquentes afin de capturer des chaînes de Markov d'ordre personnalisé. Nous avons mené une étude empirique approfondie et démontré que notre modèle surpasse les performances des différents modèles de l’état de l’art, en particulier sur des jeux de données éparses. Nous avons également intégré REBUS dans myCADservices, une plateforme collaborative de la société française Visiativ. Nous présentons notre retour d'expérience sur cette mise en production du fruit de nos travaux de recherche. Enfin, nous avons proposé une nouvelle approche pour expliquer les recommandations fournies aux utilisateurs. Le fait de pouvoir expliquer une recommandation permet de contribuer à accroître la confiance qu'un utilisateur peut avoir dans un système de recommandation. Notre approche est basée sur la découverte de sous-groupes pour fournir des explications interprétables d'une recommandation pour tous types de modèles qui utilisent comme données d’entrée les retours implicites des utilisateurs<br>Recommender systems have received a lot of attention over the past decades with the proposal of many models that take advantage of the most advanced models of Deep Learning and Machine Learning. With the automation of the collect of user actions such as purchasing of items, watching movies, clicking on hyperlinks, the data available for recommender systems is becoming more and more abundant. These data, called implicit feedback, keeps the sequential order of actions. It is in this context that sequence-aware recommender systems have emerged. Their goal is to combine user preference (long-term users' profiles) and sequential dynamics (short-term tendencies) in order to recommend next actions to a user. In this thesis, we investigate sequential recommendation that aims to predict the user's next item/action from implicit feedback. Our main contribution is REBUS, a new metric embedding model, where only items are projected to integrate and unify user preferences and sequential dynamics. To capture sequential dynamics, REBUS uses frequent sequences in order to provide personalized order Markov chains. We have carried out extensive experiments and demonstrate that our method outperforms state-of-the-art models, especially on sparse datasets. Moreover we share our experience on the implementation and the integration of REBUS in myCADservices, a collaborative platform of the French company Visiativ. We also propose methods to explain the recommendations provided by recommender systems in the research line of explainable AI that has received a lot of attention recently. Despite the ubiquity of recommender systems only few researchers have attempted to explain the recommendations according to user input. However, being able to explain a recommendation would help increase the confidence that a user can have in a recommendation system. Hence, we propose a method based on subgroup discovery that provides interpretable explanations of a recommendation for models that use implicit feedback
11
Löfström, Helena. "Time to Open the Black Box : Explaining the Predictions of Text Classification." Thesis, Högskolan i Borås, Akademin för bibliotek, information, pedagogik och IT, 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:hb:diva-14194.
Повний текст джерелаАнотація:
The purpose of this thesis has been to evaluate if a new instance based explanation method, called Automatic Instance Text Classification Explanator (AITCE), could provide researchers with insights about the predictions of automatic text classification and decision support about documents requiring human classification. Making it possible for researchers, that normally use manual classification, to cut time and money in their research, with the maintained quality. In the study, AITCE was implemented and applied to the predictions of a black box classifier. The evaluation was performed at two levels: at instance level, where a group of 3 senior researchers, that use human classification in their research, evaluated the results from AITCE from an expert view; and at model level, where a group of 24 non experts evaluated the characteristics of the classes. The evaluations indicate that AITCE produces insights about which words that most strongly affect the prediction. The research also suggests that the quality of an automatic text classification may increase through an interaction between the user and the classifier in situations with unsure predictions.
12
Reitzel-Jaffe, Deborah D. "Predicting relationship abuse, a structural equation model analysis of a social learning explanation." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 1997. http://www.collectionscanada.ca/obj/s4/f2/dsk2/ftp03/NQ28518.pdf.
Повний текст джерела
13
McGrath, Shelly Ann. "Structural Explanations, Need, Or Rurality: Predicting Availability Of Domestic Violence Emergency Services." OpenSIUC, 2009. https://opensiuc.lib.siu.edu/dissertations/53.
Повний текст джерелаАнотація:
Domestic violence occurs throughout the United States and has been cited as a major social and health problem, yet most studies have not focused on domestic violence in rural areas. In order to understand if rural women are able to receive the services they need, I propose to answer five research questions using data from Illinois: (1) Does the availability of services for victims of domestic violence vary by the degree of rurality of the county? (2) Are measures of social disorganization (poverty, racial heterogeneity, and residential stability) at the county level correlated with the availability of victim services for urban and rural counties? (3) Are measures of cultural factors at the county level correlated with the availability of victim services for urban and rural counties? (4) Is domestic violence and victim need, as reported by the Uniform Crime Reports (UCR) and as measured by orders of protection, as prevalent in rural counties as in urban counties? (5) Do measures of rurality (such as population density), social disorganization measures at the county and place level, rural cultural factors, or victim need better determine the current availability of services? The main goal of my research is to analyze whether domestic violence emergency services are available where the greatest need is located. I will analyze the Uniform Crime Reports domestic violence counts and the distribution of orders of protections from 39 counties in Illinois to analyze the societal and cultural level variables that may predict need for services and availability of services. From there I will analyze the distribution of domestic violence programs, hospitals, police, and sheriff's offices.
14
McGrath, Shelly A. "Structural explanations, need, or rurality : predicting the availability of domestic violence emergency services /." Available to subscribers only, 2009. http://proquest.umi.com/pqdweb?did=1879034001&sid=17&Fmt=2&clientId=1509&RQT=309&VName=PQD.
Повний текст джерелаАнотація:
Thesis (Ph. D.)--Southern Illinois University Carbondale, 2009.<br>"Department of Sociology." Keywords: Domestic violence, Emergency services, Rural. Includes bibliographical references (p. 202-217). Also available online.
15
Lattouf, Mouzeina. "Assessment of Predictive Models for Improving Default Settings in Streaming Services." Thesis, KTH, Skolan för kemi, bioteknologi och hälsa (CBH), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-284482.
Повний текст джерелаАнотація:
Streaming services provide different settings where customers can choose a sound and video quality based on personal preference. The majority of users never make an active choice; instead, they get a default quality setting which is chosen automatically for them based on some parameters, like internet connection quality. This thesis explores personalising the default audio setting, intending to improve the user experience. It achieves this by leveraging machine learning trained on the fraction of users that have made active choices in changing the quality setting. The assumption that user similarity in users who make an active choice can be leveraged to impact user experience was the idea behind this thesis work. It was issued to study which type of data from different categories: demographic, product and consumption is most predictive of a user's taste in sound quality. A case study was conducted to achieve the goals for this thesis. Five predictive model prototypes were trained, evaluated, compared and analysed using two different algorithms: XGBoost and Logistic Regression, and targeting two regions: Sweden and Brazil. Feature importance analysis was conducted using SHapley Additive exPlanations(SHAP), a unified framework for interpreting predictions with a game theoretic approach, and by measuring coefficient weights to determine the most predictive features. Besides exploring the feature impact, the thesis also answers how reasonable it is to generalise these models to non-selecting users by performing hypothesis testing. The project also covered bias analysis between users with and without active quality settings and how that affects the models. The models with XGBoost had higher performance. The results showed that demographic and product data had a higher impact on model predictions in both regions. Although, different regions did not have the same data features as most predictive, so there were differences observed in feature importance between regions and also between platforms. The results of hypothesis testing did not indicate a valid reason to consider the models to work for non-selective users. However, the method is negatively affected by other factors such as small changes in big datasets that impact the statistical significance. Data bias in some data features was found, which indicated a correlation but not the causation behind the patterns. The results of this thesis additionally show how machine learning can improve user experience in regards to default sound quality settings, by leveraging models on user similarity in users who have changed the sound quality to the most suitable for them.<br>Streamingtjänster erbjuder olika inställningar där kunderna kan välja ljud- och videokvalitet baserat på personliga preferenser. Majoriteten av användarna gör aldrig ett aktivt val; de tilldelas istället en standardkvalitetsinställning som väljs automatiskt baserat på vissa parametrar, som internetanslutningskvalitet. Denna avhandling undersöker anpassning av standardljudinställningen, med avsikt att förbättra användarupplevelsen. Detta uppnås genom att tillämpa maskininlärning på den andel användare som har aktivt ändrat kvalitetsinställningen. Antagandet att användarlikhet hos användare som gör ett aktivt val kan utnyttjas för att påverka användarupplevelsen var tanken bakom detta examensarbete. Det utfärdades för att studera vilken typ av data från olika kategorier: demografi, produkt och konsumtion är mest förutsägande för användarens smak i ljudkvalitet. En fallstudie genomfördes för att uppnå målen för denna avhandling. Fem prediktiva modellprototyper tränades, utvärderades, jämfördes och analyserades med två olika algoritmer: XGBoost och Logistisk Regression, och inriktade på två regioner: Sverige och Brasilien. Analys av funktionsvikt genomfördes med SHapley Additive exPlanations (SHAP), en enhetlig ram för att tolka förutsägelser med en spelteoretisk metod, och genom att mäta koefficientvikter för att bestämma de mest prediktiva funktionerna. Förutom att utforska funktionens påverkan, svarar avhandlingen också på hur rimligt det är att generalisera dessa modeller för icke-selektiva användare genom att utföra hypotesprövning. Projektet omfattade också biasanalys mellan användare med och utan aktiva kvalitetsinställningar och hur det påverkar modellerna. Modellerna med XGBoost hade högre prestanda. Resultaten visade att demografisk data och produktdata hade en högre inverkan på modellförutsägelser i båda regionerna. Däremot hade olika regioner inte samma datafunktioner som mest prediktiva, skillnader observerades i funktionsvikt mellan regioner och även mellan plattformar. Resultaten av hypotesprövningen indikerade inte på vägande anledning för att anse att modellerna skulle fungera för icke-selektiva användare. Däremot har metoden påverkats negativt av andra faktorer som små förändringar i stora datamängder som påverkar den statistiska signifikansen. Data bias hittades i vissa datafunktioner, vilket indikerade en korrelation men inte orsaken bakom mönstren. Resultaten av denna avhandling visar dessutom hur maskininlärning kan förbättra användarupplevelsen när det gäller standardinställningar för ljudkvalitet, genom att utnyttja modeller för användarlikhet hos användare som har ändrat ljudkvaliteten till det mest lämpliga för dem.
16
Hedström, Anna. "Explainable Artificial Intelligence : How to Evaluate Explanations of Deep Neural Network Predictions using the Continuity Test." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2020. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-281279.
Повний текст джерелаАнотація:
With a surging appetite to leverage deep learning models as means to enhance decisionmaking, new requirements for interpretability are set. Renewed research interest has thus been found within the machine learning community to develop explainability methods that can estimate the influence of a given input feature to the prediction made by a model. Current explainability methods of deep neural networks have nonetheless shown to be far from fault-proof and the question of how to properly evaluate explanations has largely remained unsolved. In this thesis work, we have taken a deep look into how popular, state-of-the-art explainability methods are evaluated with a specific focus on one evaluation technique that has proved particularly challenging - the continuity test. The test captures the sensitivity of explanations by measuring how an explanation, in relation to a model prediction, changes when its input is perturbed. While it might sound like a reasonable expectation that the methods behave consistently in their input domain, we show in experiments on both toy-and real-world data, on image classification tasks, that there is little to no empirical association between how explanations and networks respond to perturbed inputs. As such, we challenge the proposition that explanations and model outcomes can, and should be, compared. As a second line of work, we also point out how and why it is problematic that commonly applied ad-hoc perturbation techniques tend to produce samples that lie far from the data distribution. In the pursuit for better, more plausible image perturbations for the continuity test, we therefore present an alternative approach that relies on sampling in latent space, as learned by a probabilistic, generative model. To this end, we hope that the work presented in this thesis will not only be helpful in identifying limitations of current evaluation techniques, but that the work also contributes with ideas of how to improve them.<br>Med ett ökat intresse att använda djupa neurala nätverk i olika delar av samhället, ställs nya krav på dess tolkbarhet. Förnyat forskningsintresse har således infunnit sig för att utveckla förklaringsmodeller som kan redogöra varför ett beslut har tagits av en algoritm. Förklaringsmodellerna har dock visat sig vara långt ifrån felfria och frågan om hur man utvärderar dem har i stort sett förblivit obesvarad. I detta arbete har vi undersökt hur populära förklaringsmodeller evalueras med ett specifikt fokus på en evalueringsteknik som har visat sig vara särskilt problematisk - “kontinuitetstestet”. Testet fångar upp förklaringarnas “sensitivitet” genom att mäta hur en förklaring, i förhållande till modellens prediktion, förändras när dess input modifieras. Även om det kan låta som en rimlig förväntning att förklaringsmetoderna skall uppträda konsekvent i sin inputdomän, så visar vi i bildklassificeringsexperiment på både syntetisk och real data, att det empiriska sambandet mellan hur förklaringar och nätverk svarar på modifierad input är obetydlig. Därmed utmanar vi idén om att förklaringar och modellresultat kan och bör jämföras. Som en andra inriktning i detta arbete påpekar vi också hur och varför det är problematiskt att vanligt förekommande bildmanipuleringstekniker tenderar att producera bilder som ligger långt ifrån den ursprungliga datadistributionen. I strävan efter mer trovärdiga och bättre bildmanipuleringstekniker för kontinuitetstestet, presenterar vi därför ett alternativt tillvägagångssätt som förlitar sig på sampling i det latenta rummet, producerat av en probabilistisk generativ modell. För detta ändamål hoppas vi därför att arbetet som presenteras i denna examensarbete inte bara belyser brister i aktuella utvärderingstekniker, utan också bidrar med idéer om hur man kan förbättra dem.
17
Walsh, Emma. "Predicting the impact of health states on well-being : explanations and remedies for biased judgments." Thesis, City, University of London, 2009. http://openaccess.city.ac.uk/18257/.
Повний текст джерелаАнотація:
Affective forecasting research has demonstrated that people overestimate the impact of health states on both their own happiness and other peoples’ happiness, resulting in a disparity between a healthy sample’s predictions and the actual well-being of people living with that health state. The aim of this thesis was to explore these judgments, examine proposed explanations for the bias, and test existing and new methods for improving the accuracy. Using questionnaires, respondents predicted the impact of health states on either their own or on other peoples’ well-being. No actual difference was found in the happiness of people living and not living with health states but both groups made biased forecasts, although the predictions of respondents living with health states were biased to a lesser extent. As an explanation for inaccurate forecasts, the confound between whether a judgment was made for self happiness or others’ happiness, and whether or not the person was living with a health state, was found not to account for the bias. However, focusing too much attention on the impact of the health state, known as the focusing illusion, was concluded to be a plausible explanation. Although existing methods intended to reduce the effect of the illusion did not diminish the bias, a new method which encouraged consideration of the emotional impact of an event successfully moderated predictions. Furthermore, the bias was reduced by encouraging contemplation of the wider range of well-being of people living with health states, suggesting that biased forecasts were caused by anchoring on an extreme case. Additionally, receiving information on the happiness of people living with health states reduced the bias, but had less of an effect when presented with health state information. Thus the practicality of this remedy would be diminished in situations where health state information could not be withheld. Practical suggestions for improving affective forecasts and directions for future research are discussed.
18
Su, Susan Chih-Wen. "Female property crime offenders: Explanations from economic marginalization perspective." CSUSB ScholarWorks, 2004. https://scholarworks.lib.csusb.edu/etd-project/2673.
Повний текст джерелаАнотація:
This research explores whether women offenders who committed property crimes suffer from feminization of poverty, and social deprivations as asserted by the economic marginalization theory. Social deprivations include being a single parent with dependent children at home, being the main financial supporter of a household and being primary caretaker to minor children.
19
"Harnessing Teamwork in Networks: Prediction, Optimization, and Explanation." Doctoral diss., 2018. http://hdl.handle.net/2286/R.I.51637.
Повний текст джерелаАнотація:
abstract: Teams are increasingly indispensable to achievements in any organizations. Despite the organizations' substantial dependency on teams, fundamental knowledge about the conduct of team-enabled operations is lacking, especially at the {\it social, cognitive} and {\it information} level in relation to team performance and network dynamics. The goal of this dissertation is to create new instruments to {\it predict}, {\it optimize} and {\it explain} teams' performance in the context of composite networks (i.e., social-cognitive-information networks).
Understanding the dynamic mechanisms that drive the success of high-performing teams can provide the key insights into building the best teams and hence lift the productivity and profitability of the organizations. For this purpose, novel predictive models to forecast the long-term performance of teams ({\it point prediction}) as well as the pathway to impact ({\it trajectory prediction}) have been developed. A joint predictive model by exploring the relationship between team level and individual level performances has also been proposed.
For an existing team, it is often desirable to optimize its performance through expanding the team by bringing a new team member with certain expertise, or finding a new candidate to replace an existing under-performing member. I have developed graph kernel based performance optimization algorithms by considering both the structural matching and skill matching to solve the above enhancement scenarios. I have also worked towards real time team optimization by leveraging reinforcement learning techniques.
With the increased complexity of the machine learning models for predicting and optimizing teams, it is critical to acquire a deeper understanding of model behavior. For this purpose, I have investigated {\em explainable prediction} -- to provide explanation behind a performance prediction and {\em explainable optimization} -- to give reasons why the model recommendations are good candidates for certain enhancement scenarios.<br>Dissertation/Thesis<br>Doctoral Dissertation Computer Science 2018
20
Cornelius, Chelsea Ann. "Development of beliefs about chance and luck." Thesis, 2011. http://hdl.handle.net/2152/ETD-UT-2011-12-4900.
Повний текст джерелаАнотація:
Children ages 5 and 8 dropped a marble into a box and made predictions about which of two doors the marble would exit. Participants provided explanations and certainty ratings for each of their predictions. A lucky charm was used in a second round of the game, in which half of participants experienced an increase in success and half did not. Results indicated that older children were more cognizant of the chance nature of the game, however both age groups exhibited misconceptions about the predictability of chance outcomes. When asked to explain their overall success in Round 2, only 8 year-olds who experienced an increase in success and a perfect success rate reliably endorsed the lucky charm. Results are discussed with reference to literature on children’s and adults’ understanding of chance. We also discuss developmental patterns in the use of luck as an explanatory tool.<br>text
21
HE, XUE-FENG, and 何雪鳳. "An Application of the Prediction- Observation- Explanation Strategy to the Biochar Subject." Thesis, 2016. http://ndltd.ncl.edu.tw/handle/80444784887912805473.
Повний текст джерелаАнотація:
碩士<br>環球科技大學<br>觀光與生態旅遊系環境資源管理碩士班<br>104<br>This research uses Quasi-experimental methods to execute a biochar course for fifth grade elementary school students by experimental activities and assessment of POE (Prediction- Observation – Explanation). The aim of this research is to investigate the learning effect of the fifth grade students after the biochar teaching activity.
The Researcher directed an unequal post-test control group designed for two classes in a medium-size elementary school in Chiayi County. Class A is the experimental group and class B is the control group. Based on teaching objectives and research purposes, the researcher designed the course according to the students’ biochar learning goal and of their applying POE into biochar teaching. The data sources include before and after measured analysis, the POE learning sheet, and two-tier diagnostic test data.
According to the analysis data, the researcher comes to the following conclusions:
A. Compared to traditional didactic teaching, POE teaching establishes the concept of biochar more among the high-score students studied in this science.
B. Adopted two-tier diagnostic tests can filter and inspect whether students sank into myths or mistakes. Except for using POE teaching for the high-score students studied in science, the results of traditional didactic teaching or POE teaching for general students shows obvious discrepancy in the post test which uses two-tier diagnostic tests, and these students may need remedial teaching for myths someday
C. During the biochar classes, if a teacher doesn’t timely clarify a student's previous myths when the teaching involves the original knowledge, it may obstruct a students’ learning of new concepts.
22
Stockwell, David. "Machine learning and the problem of prediction and explanation in ecological modelling." Phd thesis, 1992. http://hdl.handle.net/1885/139553.
Повний текст джерела
23
Yang, Hui-Ting, and 楊惠婷. "Facilitating Preschoolers’ Conceptual Understanding about Gears: The Effect of the Prediction-Observation-Explanation（POE）Strategy." Thesis, 2017. http://ndltd.ncl.edu.tw/handle/8whqx2.
Повний текст джерелаАнотація:
碩士<br>國立屏東科技大學<br>幼兒保育系所<br>105<br>Many studies have found that playing building blocks can help to improve problem-solving skills and promote knowledge building. This study was conducted to examine the effects of the building blocks course embedded the prediction, observation, and explanation model (POE) on facilitating the preschoolers’ acquisition of gear concepts. The preschoolers’ alternative conceptions of gears after the treatments were also investigated. 60 preschoolers were recruited and randomly assigned into either an experimental group or a control group. The former received the building block course with the POE model embedded, the latter received the base course. The measurement included the pretest, posttest, and two-week delayed test. The course included four one-hour treatments and each participant was instructed individually during the treatment and the tests. The findings indicated that integrating the POE model into the building block course could effectively enhance the preschoolers’ acquisition of gears concepts. However, some alternative conceptions were identified after the treatments, such as “Two gears with different sizes interlock. The bigger the gear, the faster its turning speed.
24
Kuo, Chia-Yu, and 郭家諭. "Explainable Risk Prediction System for Child Abuse Event by Individual Feature Attribution and Counterfactual Explanation." Thesis, 2019. http://ndltd.ncl.edu.tw/handle/yp2nr3.
Повний текст джерелаАнотація:
碩士<br>國立交通大學<br>統計學研究所<br>107<br>There always have a trade-off: Performance or Interpretability. The complex model, such as ensemble learning can achieve outstanding prediction accuracy. However, it is not easy to interpret the complex model. Understanding why a model made a prediction help us to trust the black-box model, and also help users to make decisions. This work plans to use the techniques of explainable machine learning to develop the appropriate model for empirical data with high prediction and good interpretability.
In this study, we use the data provided by Taipei City Center for Prevention of Domestic Violence and Sexual Assault (臺北市家庭暴力暨性侵害防治中心) to develop the risk prediction model to predict the recurrence probability of violence accident for the same case before this case is resolved. This prediction model can also provide individual feature explanation and the counterfactual explanation to help social workers conduct an intervention for violence prevention.
25
Jo-Yu, Huang, and 黃若瑜. "Cognitive Style and Prior Knowledge Effect Study on Scientific Prediction, Simulation, Observation and Explanation Inquiry-Baed Learning Activity." Thesis, 2016. http://ndltd.ncl.edu.tw/handle/86542084063693838782.
Повний текст джерелаАнотація:
碩士<br>國立中央大學<br>學習與教學研究所<br>104<br>Inquiry-based learning approach plays an important role in science education, and technology has demonstrated its potential on enhancing inquiry-based learning performance. In this study,the author integrates simulation technology and POE (Prediction-Observation-Explanation)learning approach into a revised model named PSOE (Prediction-Simulation-Observation-Explanation). Sixty junior high school students were involved in this study lasting for seven weeks to evaluate the PSOE model. To provide the students a physical simulation environment,a platform, Algodoo, developed by Algoryx company was adopted in which the students can explore and observer the simulation results by manipulating the variables. Moreover, to explore the human factors affected learning performance, different cognitive style (field-dependent and field-independent students) and prior knowledge students (high prior knowledge and low prior knowledge students) were employed. The results reveal that most of the students were weak in the Explanation stage. The results also indicated that the field-independent students had higher score compared to the field-dependent student, and high-prior knowledge students had higher scores than the low-prior knowledge students, but low-prior knowledge students had much
improvement than the high-prior knowledge students. Meanwhile, the students’ science attitudes were improved after adopting PSOE inquiry-based learning activity.
26
Hochstein, Eric. "Intentionality as Methodology." Thesis, 2011. http://hdl.handle.net/10012/6530.
Повний текст джерелаАнотація:
In this dissertation, I examine the role that intentional descriptions play in our scientific study of the mind. Behavioural scientists often use intentional language in their characterization of cognitive systems, making reference to “beliefs”, “representations”, or “states of information”. What is the scientific value gained from employing such intentional terminology?
I begin the dissertation by contrasting intentional descriptions with mechanistic descriptions, as these are the descriptions most commonly used to provide explanations in the behavioural sciences. I then examine the way that intentional descriptions are employed in various scientific contexts. I conclude that while mechanistic descriptions characterize the underlying structure of systems, intentional descriptions allow us to generate predictions of systems while remaining agnostic as to their mechanistic underpinnings.
Having established this, I then argue that intentional descriptions share much in common with statistical models in the way they characterize systems. Given these similarities, I theorize that intentional descriptions are employed within scientific practice as a particular type of phenomenological model. Phenomenological models are used to study, characterize, and predict the phenomena produced by mechanistic systems without describing their underlying structure. I demonstrate why such models are integral to our scientific discovery, and understanding, of the mechanisms that make up the brain.
With my account on the table, I then look back at previous accounts of intentional language that philosophers have offered in the past. I highlight insights that each brought to our understanding of intentional language, and point out where each ultimately goes astray.
I conclude the dissertation by examining the ontological implications of my theory. I demonstrate that my account is compatible with versions of both realism, and anti-realism, regarding the existence of intentional states.
27
KU, PING, and 谷冰. "The Comparison of Design Experimental Course and Micro Teaching of Prediction - Observation - More Explanation Teaching Strategy by Different Background Graduate Students." Thesis, 2018. http://ndltd.ncl.edu.tw/handle/3z467m.
Повний текст джерелаАнотація:
碩士<br>國立臺北教育大學<br>自然科學教育學系<br>106<br>This study with adopted mixed methods of quality and quantity research explores the effect of Prediction - Observation - more Explanation (PO+E) method learned by graduate students of different backgrounds on the ability of experiment design, micro teaching, scientific explanation, and scientific inquiry. 14 graduate students (including 7 in-service teachers and 7 non-service teachers) taking the course of Advanced Entomology in the Department of Science Education were selected. During the course, graduate students first study the general concept of entomology and PO+E and observe teaching demonstrations, and then progress into the first stage of imitation experiment and micro teaching and the second stage of self-designed experiments and micro teaching. The researcher performs the analysis with the collection of the data of PO+E study sheets and written interviews, combined with designed experiment evaluation form and scale, Micro Teaching Assessment of PO+E (MTAP), Assessment of Scientific Interpretation ability (ASI), and Science Inquiry Competence Test (SICT). The research result indicated that after completing this course, graduate students can utilize PO+E teaching strategy with great obvious progress on the ability of designed experiments, micro-teaching, scientific explanation, and scientific inquiry.The comparison between in-service teacher group and non-service teacher group is as follows: First in the designed experiment with PO+E explanations in-service teachers scored higher but the non-service teachers made substantial progress; secondly, in the field of micro-teaching, the in-service teachers are superior to the non-service teachers in terms of scoring and progress; third, the ability of scientific interpretation can be gradually growing, the in-service teachers scored higher, but the non-service teachers has made great progress; fourth, in the scientific inquiry ability aspect, the in-service teachers’ score is high, but the non-service teachers have a large progress. Moreover there is no statistically significant difference in the scientific inquiry ability between in-service teachers and non-in-service teachers before or after class. In a word, graduate students with different backgrounds learning the PO+E teaching method can improve their ability of experiment design, micro-teaching, scientific explanation ability, and scientific inquiry ability. The results of this study can be used for reference in future research.
28
HUANG, YI-SHENG, and 黃奕升. "Elementary Pre-Service Teachers Apply a Prediction-Observation-More Explanation to Design Science Lesson Plans and Teach in the Elementary School." Thesis, 2019. http://ndltd.ncl.edu.tw/handle/wpbnc9.
Повний текст джерелаАнотація:
碩士<br>國立臺北教育大學<br>自然科學教育學系<br>107<br>This study supported 24 elementary pre-service teachers to design science lesson plans using a Prediction-Observation-more Explanation (PO+E Teaching method), and the "Natural science teaching materials and methods" course in the Department of Science Education provided the training. A mixed method research design was used comprising Quantitative and qualitative research. First, elementary pre-service teachers were instructed in using the PO+E Teaching method and watched case films to see how it could be used (3 weeks). Second, elementary pre-service teachers were divided into eight groups to discuss and plan how to use the PO+E Teaching method to design science lessons (4 weeks). Then, elementary pre-service teachers implemented micro-teaching using the PO+E Teaching method in the class, and the professor and peers provided comments and feedback (4 weeks). Finally, elementary pre-service teachers taught their science lessons in elementary school classrooms, and counselors evaluated their teaching. It took pretest and posttest on elementary pre-service teachers with Science Inquiry Competence Test (SICT) at the course start and the end. SPSS was used to analyses the SICT quantitative data, and the collected science lesson plans, learning sheets, and Micro Teaching Assessment of PO+E (MTAP) were analyzed with Hermeneutic method and Kappa.
The results were as follows:
1.All elementary pre-service teachers were able to use the PO+E Teaching method to design science lessons, they achieved good degree.
2.The pre-test and post-test SICT results showed significant difference (p < .05), indicating that elementary pre-service teachers had improved their science inquiry competence skills after completing the course.
3.Qualitative data showed that elementary pre-service teachers familiar with science teaching with the PO+E Teaching method.
4.The PO+E Teaching method was effective in supporting elementary pre-service teachers to design science lesson plans.
29
"Predicting and Interpreting Students Performance using Supervised Learning and Shapley Additive Explanations." Master's thesis, 2019. http://hdl.handle.net/2286/R.I.53452.
Повний текст джерелаАнотація:
abstract: Due to large data resources generated by online educational applications, Educational Data Mining (EDM) has improved learning effects in different ways: Students Visualization, Recommendations for students, Students Modeling, Grouping Students, etc. A lot of programming assignments have the features like automating submissions, examining the test cases to verify the correctness, but limited studies compared different statistical techniques with latest frameworks, and interpreted models in a unified approach.
In this thesis, several data mining algorithms have been applied to analyze students’ code assignment submission data from a real classroom study. The goal of this work is to explore
and predict students’ performances. Multiple machine learning models and the model accuracy were evaluated based on the Shapley Additive Explanation.
The Cross-Validation shows the Gradient Boosting Decision Tree has the best precision 85.93% with average 82.90%. Features like Component grade, Due Date, Submission Times have higher impact than others. Baseline model received lower precision due to lack of non-linear fitting.<br>Dissertation/Thesis<br>Masters Thesis Computer Science 2019