ResearchWorks
https://digital.lib.washington.edu:443/researchworks
The ResearchWorks digital repository system captures, stores, indexes, preserves, and distributes digital research material.Wed, 25 Apr 2018 08:09:38 GMT2018-04-25T08:09:38Z￼An Investigation of Evaluation Methods of Weighted-Point Landscaping Policy - in Seattle and ￼Malmö
http://hdl.handle.net/1773/41848
￼An Investigation of Evaluation Methods of Weighted-Point Landscaping Policy - in Seattle and ￼Malmö
Hoff, Josh
Urban vegetation plays an important role in influencing ecological function beyond the individual plant scale. City policies set required quantity and quality parameters for urban vegetation that determine the environmental and societal outcomes and ultimately affect urban residents’ wellbeing. The way urban vegetation policies are structured varies due to context but one structure that is primed for analysis, due to its clear outcome objectives and transparent structure, is the use of a weighted-point system landscape policy. The objective of this thesis is to test the applicability of a methodological evaluation of weighted-point landscaping policy by inventorying in two locations Seattle, Washington and Malmö, Sweden, the quantity and quality of vegetated outcomes at different points of time. The weighed-point landscaping policy is a structure containing a menu of obligatory landscaping options to fulfill defined requirements organized by point calculations for built environment, applies each intended to produce specific urban vegetation outcomes. Despite the intention, outcomes from urban vegetation policy vary in implementation and maintenance over time. Evaluation of the required outcomes hold relevant stakeholders accountable by inventorying fulfillment and deviation from policy intentions. Evaluative methodology and methods are tested to formalize the cyclical relationship between actual policy outcomes and feedback to improve the policy. The feedback from these methods have the potential to inform policy success over time. Two methods, a supervised Object Based Image Analysis (OBIA) and a vertical structure inventory, have the promise to be effective measurement tools given carefully chosen considerations and inputs. Applying these methods has the potential to inform a modified-BAR vegetation indicator useful for comprehensively referencing the quantity and quality of urban vegetation outcomes from weighted-point landscaping policy. In this thesis, the referenced two methods are used to analyze projects across three time periods and two weighted-point landscaping policy contexts. Lessons learned from this analysis provides a preliminary frame work for generating feedback of the policy’s outcomes to inform policy makers and relevant stakeholders.
Thesis (Master's)--University of Washington, 2018
http://hdl.handle.net/1773/41848Coevolution Regression and Composite Likelihood Estimation for Social Networks
http://hdl.handle.net/1773/41847
Coevolution Regression and Composite Likelihood Estimation for Social Networks
He, Yanjun
We study how social networks and nodal attributes influence each other over time. A multiplicative coevolution regression (MCR) model is proposed for longitudinal network and nodal attribute data. The coevolution model is based on the following three principles: autocorrelation, homophily and contagion. For the Gaussian MCR model, the maximum likelihood estimates can be obtained using ordinary least squares. We also extend the Gaussian MCR so that it can include latent factors or model ordinal data. A Bayesian method using Markov Chain Monte Carlo (MCMC) is used to estimate the parameters and latent factors. We then focus on developing a scalable method to estimate the parameters in models of very large binary network datasets. Maximum likelihood estimates are generally impossible to obtain because the full likelihood involves an intractable high dimensional integral. Also, full-likelihood Bayesian estimation is impractical for very large datasets as the MCMC algorithm is very slow. We propose a triadic composite likelihood estimation method for exchangeable latent Gaussian network models, and extend it to q-node composite likelihood estimation for other exchangeable and non-exchangeable models. The maximum composite likelihood estimates are obtained by optimizing the composite likelihood using a stochastic gradient-based algorithm, where the gradients are approximated using Monte Carlo samples. For networks of moderate size, we show via simulations that composite likelihood estimation provides estimates as accurate as those provided by fully Bayesian estimation using MCMC. For very large datasets, fully Bayesian estimation is impractical, but composite likelihood estimation is feasible as its computational cost is essentially constant as a function of the network size.
Thesis (Ph.D.)--University of Washington, 2018
http://hdl.handle.net/1773/41847Linear Structural Equation Models with Non-Gaussian Errors: Estimation and Discovery
http://hdl.handle.net/1773/41846
Linear Structural Equation Models with Non-Gaussian Errors: Estimation and Discovery
Wang, Y. Samuel
Linear structural equation models (SEMs) are multivariate models which encode direct causal effects. We focus on SEMs in which unobserved latent variables have been marginalized and only observed variables are explicitly modeled. In this thesis, we study three problems where the distribution of the stochastic errors in the SEMs, and thus the corresponding data, are non-Gaussian. Throughout, we utilize graphical models to represent the causal structure. First, we consider estimation of model parameters using an empirical likelihood framework when the causal structure is known. Asymptotically, under very mild conditions on the error distributions, this approach yields normal estimators and well calibrated confidence intervals and hypothesis tests. However, the procedure can be computationally expensive and suffer from poor performance when the sample size is small. We propose several modifications to a naive procedure and show that empirical likelihood can be an attractive alternative to existing methods when the data is non-Gaussian. The models considered in this section correspond to general mixed graphs. We then consider the problem of estimating the underlying structure. Most of the previous work on causal discovery focuses on estimating an equivalence class of graphs rather than a specific graph. However, Shimizu et al. (2016) show that under certain conditions, when the errors are non-Gaussian, the exact causal structure can be identified. We extend these results in two ways. In Chapter 3, we show that when there is no unobserved confounding and the causal structure is suitably sparse, the identification results can be extended to the high-dimensional setting where the number of variables exceed the number of observations. The models considered correspond to directed acyclic graphs (DAGs) with bounded in-degree. In Chapter 4, we show that non-Gaussian errors also allow for identification of the specific graph when unobserved confounding occurs in a restricted way. In particular, we consider the case where the underlying model corresponds to a bow-free acyclic path diagram (BAP). The proposed method consistently estimates the underlying structure, and unlike previous results does not require the number of latent variables or distribution of the errors to be specified in advance.
Thesis (Ph.D.)--University of Washington, 2018
http://hdl.handle.net/1773/41846Brown Induction and Red/Green Hue Shifts
http://hdl.handle.net/1773/41845
Brown Induction and Red/Green Hue Shifts
DeLawyer, Tanner Jonathan
This body of work details a number of experiments relating to brown induction (the change of appearance of a stimulus from yellow to brown) and shifts in red/green hue balance (the balance point of stimuli across the yellow-blue spectrum where they appear neither reddish nor greenish). These experiment involve variations of targets and their surrounding stimuli in the dimensions of luminance, saturation, and hue, as well as variations in the manner of their optical presentation (monocular [stimulus only in one eye], binocular [same stimulus both eyes], or dichoptic [different stimuli in each eye] presentations). It was found that brown induction is strongest for stimuli that are sufficiently darker than their surround and slightly desaturated. Although brown stimuli show a systematic shift in their red/green balance compared to yellow stimuli this appears to be controlled by a separate mechanism that is present for all targets that vary along the yellow-blue spectrum (including achromatic gray targets) when they are darker than their surrounds. It was also found that brown induction can occur both prior to and after the cortical combination of signals from the two eyes (in both monocular and binocular pathways, respectively), showing an enhancement from perceptually contiguous bright surrounds presented in either the same or opposite eye as a target. No similar effect was observed for red/green balances, which appear to be influenced only in monocular pathways and have a different relationship with surround contiguity. There are also differences in the effects of target and surround size on brown induction compared to red/green balance shifts. Brown represents a categorical hue change from yellow that is usually accompanied by a red/green balance shift, but such red/green balance shifts can also occur independently of a categorical hue changes as seen with achromatic gray stimuli. These many differences between brown induction and red/green balance shifts suggest that they are two separate phenomena that can co-occur but are not directly related, suggesting different neural mechanisms for each. In addition to academic significance, understanding and accurately modeling these effects is increasingly important for issues related to both human and machine vision.
Thesis (Ph.D.)--University of Washington, 2018
http://hdl.handle.net/1773/41845