Spelling suggestions: "subject:"No-U-Turn sampler"" "subject:"No-U-Turn ampler""
1 |
Bayesian Estimation of Mixture IRT Models using NUTSAl Hakmani, Rahab 01 December 2018 (has links)
The No-U-Turn Sampler (NUTS) is a relatively new Markov chain Monte Carlo (MCMC) algorithm that avoids the random walk behavior that common MCMC algorithms such as Gibbs sampling or Metropolis Hastings usually exhibit. Given the fact that NUTS can efficiently explore the entire space of the target distribution, the sampler converges to high-dimensional target distributions more quickly than other MCMC algorithms and is hence less computational expensive. The focus of this study is on applying NUTS to one of the complex IRT models, specifically the two-parameter mixture IRT (Mix2PL) model, and further to examine its performance in estimating model parameters when sample size, test length, and number of latent classes are manipulated. The results indicate that overall, NUTS performs well in recovering model parameters. However, the recovery of the class membership of individual persons is not satisfactory for the three-class conditions. Also, the results indicate that WAIC performs better than LOO in recovering the number of latent classes, in terms of the proportion of the time the correct model was selected as the best fitting model. However, when the effective number of parameters was also considered in selecting the best fitting model, both fully Bayesian fit indices perform equally well. In addition, the results suggest that when multiple latent classes exist, using either fully Bayesian fit indices (WAIC or LOO) would not select the conventional IRT model. On the other hand, when all examinees came from a single unified population, fitting MixIRT models using NUTS causes problems in convergence.
|
2 |
An application of Bayesian Hidden Markov Models to explore traffic flow conditions in an urban areaAndersson, Lovisa January 2019 (has links)
This study employs Bayesian Hidden Markov Models as method to explore vehicle traffic flow conditions in an urban area in Stockholm, based on sensor data from separate road positions. Inter-arrival times are used as the observed sequences. These sequences of inter-arrival times are assumed to be generated from the distributions of four different (and hidden) traffic flow states; nightly free flow, free flow, mixture and congestion. The filtered and smoothed probability distributions of the hidden states and the most probable state sequences are obtained by using the forward, forward-backward and Viterbi algorithms. The No-U-Turn sampler is used to sample from the posterior distributions of all unknown parameters. The obtained results show in a satisfactory way that the Hidden Markov Models can detect different traffic flow conditions. Some of the models have problems with divergence, but the obtained results from those models still show satisfactory results. In fact, two of the models that converged seemed to overestimate the presence of congested traffic and all the models that not converged seem to do adequate estimations of the probability of being in a congested state. Since the interest of this study lies in estimating the current traffic flow condition, and not in doing parameter inference, the model choice of Bayesian Hidden Markov Models is satisfactory. Due to the unsupervised nature of the problematization of this study, it is difficult to evaluate the accuracy of the results. However, a model with simulated data and known states was also implemented, which resulted in a high classification accuracy. This indicates that the choice of Hidden Markov Models is a good model choice for estimating traffic flow conditions.
|
3 |
Evaluating Markov Chain Monte Carlo Methods for Estimating Systemic Risk Measures Using Vine Copulas / Utvärdering av Markov Chain Monte Carlo-metoder vid estimering av systemisk risk under portföljmodellering baserad på Vine CopulasGuterstam, Rasmus, Trojenborg, Vidar January 2021 (has links)
This thesis attempts to evaluate the Markov Chain Monte Carlo (MCMC) methods Metropolis-Hastings (MH) and No-U-Turn Sampler (NUTS) to estimate systemic risk measures. The subject of analysis is an equity portfolio provided by a Nordic asset management firm, which is modelled using a vine copula. The evaluation considers three different crisis outcomes on a portfolio level, and the results are compared with a Monte Carlo (MC) benchmark. The MCMC samplers attempt to increase sampling efficiency by sampling from these crisis events directly, which is impossible for an MC sampler. The resulting systemic risk measures are evaluated both on the portfolio level as well as marginal level. The results are divided. In part, the MCMC samplers proved to be efficient in terms of accepted samples, where NUTS outperformed MH. However, due to the practical implementation of the MCMC samplers and the vine copula model, the computational time required outweighed the gains in sampler efficiency - causing the MC sampler to outperform both MCMC samplers in certain settings. For NUTS, there seems to be great potential in the context of estimating systemic risk measures as it explores high-dimensional and multimodal joint distributions efficiently with low autocorrelation. It is concluded that asset management companies can benefit from both using vine copulas to model portfolio risk, as well as using MC or MCMC methods for evaluating systemic risk. However, for the MCMC samplers to be of practical relevance, it is recommended to further investigate efficient implementations of vine copulas in the context of MCMC sampling. / Detta examensarbete utvärderar Markov Chain Monte Carlo (MCMC)-metoderna No-U-Turn Sampler (NUTS) och Metropolis-Hastings (MH) vid uppskattning av systemiska riskmått. För att göra detta används en vine copula för att modellera en portfölj, baserad på empirisk data från ett nordiskt kapitalförvaltningsföretag. Metoderna utvärderas givet tre olika krishändelser och jämförs därefter med ett Monte Carlo (MC) benchmark. MCMC-metoderna försöker öka samplingseffektiviteten genom att simulera direkt från dessa krishändelser, vilket är omöjligt för en klassisk MC-metod. De resulterande systemiska riskmåtten utvärderas både på portföljnivå och på marginalnivå. Resultaten är delade. Dels visade sig MCMC-metoderna vara effektiva när det gäller accepterade samples där NUTS överträffade MH. Dock, med anledning av av den praktiska implementationen av MCMC-metoderna och vine copula modellen var beräkningstiden för hög trots effektiviteten hos metoden - vilket fick MC-metoden att överträffa de andra metoderna i givet dessa särskilda kontexter. När det kommer till att uppskatta systemiska riskmått finns det dock stor potential för NUTS eftersom metoden utforskar högdimensionella och multimodala sannolikhetsfördelningar effektivt med låg autokorrelation. Vi drar även slutsatsen att kapitalförvaltare kan dra nytta av att både använda riskmodeller baserade på vine copulas, samt använda MC- eller MCMC-metoder för att utvärdera systemisk risk. För att MCMC-metoderna ska vara av praktisk relevans rekommenderas det dock att framtida forskning görs där mer effektiva implementeringar av vine copula-baserade modeller görs i samband med MCMC-sampling.
|
Page generated in 0.0377 seconds