Semi-structured interview protocol for constructing logic models

This paper details a semi-structured interview protocol that evaluators can use to develop a logic model of a program's services and outcomes. The protocol presents a series of questions, which evaluators can ask of specific program informants, that are designed to: (1) identify key informants basic background and contextual information, (2) generate logic model elements, (3) model program inputs, activities, outputs, and outcomes, (4) build a rational theory, (5) develop a program theory, (6) prioritize logic model elements, and (7) build a graphical or tabular logic model. The paper will also provide an example of how this approach was used to develop a logic model for a youth mentoring program. It is our hope and belief that with this interview protocol, novice evaluators will be able to generate comprehensive logic models like seasoned professional evaluators.     

Checking validity of monotone domain mean estimators

Estimates of population characteristics such as domain means are often expected to follow monotonicity assumptions. Recently, a method to adaptively pool neighbouring domains was proposed, which ensures that the resulting domain mean estimates follow monotone constraints. The method leads to asymptotically valid estimation and inference, and can lead to substantial improvements in efficiency, in comparison with unconstrained domain estimators. However, assuming incorrect shape constraints may lead to biased estimators. Here, we develop the Cone Information Criterion for Survey Data as a diagnostic method to measure monotonicity departures on population domain means. We show that the criterion leads to a consistent methodology that makes an asymptotically correct decision choosing between unconstrained and constrained domain mean estimators. The Canadian Journal of Statistics 47: 315–331; 2019 © 2019 Statistical Society of Canada     

A fast algorithm for non-negativity model selection

An efficient optimization algorithm for identifying the best least squares regression model under the condition of non-negative coefficients is proposed. The algorithm exposits an innovative solution via the unrestricted least squares and is based on the regression tree and branch-and-bound techniques for computing the best subset regression. The aim is to filling a gap in computationally tractable solutions to the non-negative least squares problem and model selection. The proposed method is illustrated with a real dataset. Experimental results on real and artificial random datasets confirm the computational efficacy of the new strategy and demonstrates its ability to solve large model selection problems that are subject to non-negativity constrains.     

Do small cities affect bird assemblages? An evaluation from Patagonia

Urban Ecosystems - Most studies about the impact of urbanization on bird assemblages are from the Northern Hemisphere and most previous research has reported that cities reduced bird species...     

Estimating Traffic Emissions Using Demographic and Socio-Economic Variables in 18 Chilean Urban Areas

A macro-scale methodology for vehicle emissions estimation is described. The methodology is based on both correlations between activity level and PM, CO, THC and NO _x vehicle emissions and relationships between demographic and socio-economic variables and transportation activity level. First, pollutant emissions were correlated with transportation activity, expressed as vehicle-km/year, using existing data collected from mobile sources emission inventories in nine urban cities of Chile. Second, demographic and socio-economic variables were pre-selected from those that could intuitively be correlated with vehicle activity level and considering the data availability. Using the individual R ² correlation coefficient as variable selection criterion, population, the number of vehicles, fuel consumption, gross domestic product, average family incomes and road kilometers were finally chosen. A different set of explicative variables was considered for different vehicle categories, based on the selection criterion above mentioned. Then, correlation functions between these variables and transport activity were obtained by non-linear Gauss–Newton least square method. This methodology was applied to eighteen provinces of the country obtaining total annual emission for mobile sources, divided into six main vehicles categories.     

Estimating the Probability of Default with Applications in Provisioning the Portfolio of Clients of a Credit Institution

In this paper we study the implications of estimating the expected probabilities of default (PDs), for the firms that are already part of a credit portfolio of a bank (or in general, of a credit institution), on the provisioning process of that portfolio. An applied example of an estimation of PDs is conducted on a representative sample of companies from a real portfolio of an important bank in Romania (among top five banks as size). With respect to what is recommended by the standard practice and central bank authorities, our provisioning application (based on the logistic estimation of PDs) results in a higher risk aversion for those credit classes of superior financial reliability, respectively leads to a risk underestimation for the companies in the credit classes with low levels of reliability. Besides, the results of our econometric exercise of estimating PDs reveal the importance of the solvency as predictor of the default probability, as well as the negative effect of the services sector, in the context of the financial and economic Romanian crisis of the year 2008.     

Regression models for functional data by reproducing kernel Hilbert spaces methods

Non-parametric regression models are developed when the predictor is a function-valued random variable X={_Xt}t∈T$X=\{X_t{\}}_{t\in T}$. Based on a representation of the regression function f(X)$f(X)$ in a reproducing kernel Hilbert space such models generalize the classical setting used in statistical learning theory. Two applications corresponding to scalar and categorical response random variable are performed on stock-exchange and medical data. The results of different regression models are compared.     

Determining an Optimal Penetration Among Weighted Regions in Two and Three Dimensions

We present efficient algorithms for solving the problem of computing an optimal penetration (a ray or a semi-ray) among weighted regions in 2-D and 3-D spaces. This problem finds applications in several areas, such as radiation therapy, geological exploration, and environmental engineering. Our algorithms are based on a combination of geometric techniques and optimization methods. Our geometric analysis shows that the d-D (d = 2, 3) optimal penetration problem can be reduced to solving O(n ^2(d–1)) instances of certain special types of non-linear optimization problems, where n is the total number of vertices of the regions. We also give implementation results of our 2-D algorithms.