1,721,210 research outputs found
Anvendelse av Bayesiansk kalibrering for forplantning av usikkerhet i dynamiske modeller
Full text linkThe thesis is about quantification of uncertainties in complex models. Models are built to describe, explain or predict a real world outcome. It is well known that models are related with uncertainty, and that uncertainties are related to how close the simulation is to the real world outcome. Still, uncertainties are rarely quantified in dynamic models. We have focused on parameter uncertainty and output uncertainty derived from the parameters. Uncertainty originated from the empirical data is integrated into the posterior parameter distributions through the likelihood functions. Additionally, uncertainty related to the representativeness of the collected data to the population has been focused.
The Bayesian statistical framework, with the Markov chain Monte Carlo algorithm random walk Metropolis was used for model calibration in the four papers. The algorithm was found simple in idea and implementation into the computer program Matlab, but challenges emerged when the method was used at complex models. In this work these challenges have been pursued together with searching for efficiency improvements in order to make as few model evaluations as possible.
Paper I: explores the challenges emerging when applying Bayesian calibration to a complex deterministic dynamic model of snow depth. How prior information and new data affect the calibration process, the parameter estimates and model outputs were demonstrated. Parameter uncertainty and model uncertainty derived from the parameters were quantified, visualized and assessed. The random walk Metropolis algorithm was used and in order to reach convergence more effectively, informative priors, Sivias’ likelihood, reflection at the prior boundaries and updating the proposal distribution with parts of the data gave successful results. Methods for objective and correct determination of Markov chain convergence were studied, and the use of multiple chains and the Gelman-Rubin method was found useful.
Paper II: presents a dynamic model for snow cover, soil frost and surface ice. The Bayesian approach was used for model calibration and sensitivity analysis identified the non-important parameters.
Paper III: shows the importance of splitting the data several times in two for model development and assessment/selection, for the model to fit well to novel data from the system and not only to the specific data at hand. Different models of ascospore maturity of Venturia inaequalis were further developed and compared by the deviance information criterion and root mean square error of prediction to show model improvements, and the analysis of variance was used to show significance of the improvements.
Paper IV: examines the potential effects of selection of likelihood function when calibration a model. Since the likelihood function is rarely known for certain, but gives a reasonable quantification of how probable the data are given model outcome, it is of great importance to quantify the effect of using different likelihood functions on parameter uncertainty and on model output uncertainty derived from the parameters.Denne avhandlingen omhandler kvantifisering av usikkerhet i komplekse modeller. Modeller bygges for å beskrive, forklare og predikere virkelige systemer. Selv om det er kjent at usikkerhet er knyttet til modeller og dermed er relatert til hvor lik den simulerte og den virkelige verdien er, blir usikkerhet sjeldent kvantifisert i dynamiske modeller. Det har i denne avhandlingen blitt lagt vekt på parameterusikkerhet og usikkerhet i utgangsdata fra modeller med opprinnelse i parametrene og empiriske data. I tillegg har det blitt fokusert på usikkerhet i forhold til hvor godt de innsamlede observasjonene representerer populasjonen.
Bayesiansk statistikk har blitt anvendt til å kalibrere modellene i de fire påfølgende artiklene ved hjelp av Markov chain Monte Carlo algoritmen random walk Metropolis. Algoritmen er enkel å forstå og å implementere i dataprogrammet Matlab, men utfordringer oppstod ved anvendelse på komplekse modeller. Disse utfordringene og effektivitetsforbedringer ved å minimere antall modellevalueringer har blitt vektlagt.
Artikkel I: utforsker utfordringer ved bruk av Bayesiansk kalibrering på en kompleks deterministisk dynamisk modell for snødybde. Fokus er lagt på hvordan den opprinnelige usikkerheten, a’priori usikkerheten, sammen med nye innsamlede data gjennom rimelighetsfunksjonen påvirker kalibreringsprosessen, parameter estimatene og modellens utgangsdata. Parameterusikkerhet og modellusikkerhet grunnet parametrene ble kvantifisert, vist og vurdert. Random walk Metropolis algoritmen ble anvendt, og for å oppnå konvergens raskere ble informative fordelinger på parametrene, Sivias’ rimelighetsfunksjon, refleksjon og å oppdatere forslagsfordelingen med deler av dataene testet med gode resultater. Det ble dessuten lagt vekt på viktigheten av en metode for å avgjøre både objektivt og korrekt når kjedene konvergerte, hvor parallelle kjeder og Gelman-Rubins metode ble funnet nyttig.
Artikkel II: presenterer en dynamisk modell for snødybde, frostdybde og overflate-is. Bayesiansk rammeverk ble anvendt for å kalibrere modellen og sensitivitetsanalyse identifiserte de mindre viktige parametrene.
Artikkel III: viser hvor viktig det er å dele data flere ganger i to for modell utvikling og modell validering for at modellen ikke kun skal passe de spesifikke dataene, men også nye data fra det samme systemet. Ulike modeller for sporemodning av Venturia inaequalus ble videreutviklet og sammenlignet ved bruk av kriteriene devianse informasjons kriteri (DIC) og prediksjonsfeil (RMSEP) for å vise modellforbedringer. Variansanalyse ble anvendt for å angi statistisk signifikans til forbedringene.
Artikkel IV: undersøker effekten av rimelighetsfunksjonen på en snødybdemodell. Siden rimelighetsfunksjonen sjelden er kjent, men kun gir en fornuftig kvantifisering av hvor sannsynlig data er gitt modellens utdata, er det viktig å kvantifisere effekten av å anvende ulike rimelighetsfunksjoner på parameterusikkerhet og på usikkerheten relatert til modellens utgangsdata med opprinnelse i apriori parameterusikkerhet og empiriske data.Biofors
Mobile technology for climate change adaptation in agriculture
Full text linkClimaRice II is exploring the potential for use of mobile technologies in the context of climate change adaptation in agriculture. Modern mobile telephone technology is a key component of the ongoing communication revolution which in turn has great potentials for social change and development. The Indian telecommunication industry is the world's fastest growing industry with 811.59 million mobile phone subscribers as of March 2011. Most farmers are already using mobile phones for various day to day needs, but the technology has a wider potential in supporting their main profession; agriculture. Linking mobile technology with adaptation measures developed in ClimaRice projects could form new and powerful measures to meet the threats from climate change and provide support in sustaining rice production.publishedVersio
Going Beyond Counting First Authors in Author Co-citation Analysis
Full text linkThe present study examines one of the fundamental aspects of author co-citation analysis (ACA) - the way co-citation
counts are defined. Co-citation counting provides the data on which all subsequent statistical analyses and mappings
are based, and we compare ACA results based on two different types of co-citation counting - the traditional type that
only counts the first one among a cited work's authors on the one hand and a non-traditional type that takes into
account the first 5 authors of a cited work on the other hand. Results indicate that the picture produced through this non-traditional author co-citation counting contains more coherent author groups and is therefore considerably clearer. However, this picture represents fewer specialties in the research field being studied than that produced through the traditional first-author co-citation counting when the same number of top-ranked authors is selected and analyzed. Reasons for these effects are discussed
Variations on the Author
Full text link“Variations on the Author” discusses two of Eduardo Coutinho’s recent films (Um Dia na Vida, from 2010, and Últimas Conversas, posthumously released in 2015) and their contribution to the general question of documentary authorship. The director’s filmography is characterized by a consistent yet self-effacing form of authorial self-inscription: Coutinho often features as an interviewer that rather than express opinions propels discourses; an interviewer that is good at listening. This mode of self-inscription characterizes him as an author who is not expressive but who is nonetheless markedly present on the screen. In Um Dia na Vida, however, Coutinho is completely absent form the image, while Últimas Conversas, on the contrary, includes a confessional prologue that moves the director from the margins to the center of his films. This article examines the ways in which these works stand out in the filmography of a director who offers new insights into the notion of cinematic authorship
Appropriate Similarity Measures for Author Cocitation Analysis
Full text linkWe provide a number of new insights into the methodological discussion about author cocitation analysis. We first argue that the use of the Pearson correlation for measuring the similarity between authors’ cocitation profiles is not very satisfactory. We then discuss what kind of similarity measures may be used as an alternative to the Pearson correlation. We consider three similarity measures in particular. One is the well-known cosine. The other two similarity measures have not been used before in the bibliometric literature. Finally, we show by means of an example that our findings have a high practical relevance.information science;Pearson correlation;cosine;similarity measure;author cocitation analysis
Tidlige prognoser for kornavlingene ved bruk av værdata - Sluttrapport
Full text linkTidlige og presise prognoser over norsk kornproduksjon er viktig for en god og effektiv regulering av kornmarkedet. Målet med dette prosjektet har vært å utvikle en ny og objektiv metode for å lage tidlige (per 1. august) prognoser på dekaravlinger (kg korn per daa), ved å kombinere historiske sammenhenger mellom vær og avlinger. De værbaserte prognosene for bygg hadde høyst 10 % avvik fra målingene, og avviket var mindre enn 3 % i 2005 og 2006. Dermed var metoden bedre enn dagens metode i to av tre år testet. Avviket i havreprognosene overskred ikke 7 % for den nye metoden, og var som bygg bedre enn dagens metode i to av tre år. For rug var avviket på linje med tilsvarende for havre (<7 %), men her var dagens metode noe bedre i 2006 og 2007. For hvete traff den nye metoden godt i 2005 og 2006 (<7 %), og bedre enn dagens metode i 2005, men marginalt dårligere i 2006. Begge metodene overestimerte hveteavlingene kraftig i 2007.publishedVersio
Været i vekstsesongen 2010
Full text linkDette er en kort beskrivelse av været i vekstsesongen 2010 basert på data registrert ved Bioforsks klimastasjoner og noen detaljer hentet fra klimatologiske oversikter utgitt av Meteorologisk institutt.publishedVersio
Fire blight in Norway - An assessment of the plant health risk for the plant disease fire blight in Norway
Full text linkDispelling the Myths Behind First-author Citation Counts
Full text linkWe conducted a full-scale evaluative citation analysis study of scholars in the XML research field to explore just how different from each other author rankings resulting from different citation counting methods actually are, and to demonstrate the capability of emerging data and tools on the Web in supporting more realistic citation counting methods. Our results contest some common arguments for the continued
use of first-author citation counts in the evaluation of scholars, such as high correlations between author rankings by first-author citation counts and other citation
counting methods, and high costs of using more realistic citation counting methods that are not well-supported by the ISI databases. It is argued that increasingly available digital full text research papers make it possible for citation analysis studies to go beyond what the ISI databases have directly supported and to employ more
sophisticated methods
- …
