Small Area Estimation of the Homeless in Los Angeles: An Application of Cost-Sensitive stochastic Gradient Boosting
By Brian Kriegler, Richard A. Berk. 2010
50 citations recorded by the source corpus.
In many metropolitan areas, eorts are made to count the homeless to ensure proper provision of social services. Some areas are very large, which makes spatial sampling a viable alternative to an enumeration of the entire terrain. Counts are observed in sampled regions but must be imputed in unvisited areas. Along with the imputation process, the costs of underestimating and overestimating may be dierent. For example, if precise estimation in areas with large homeless counts is critical, then underestimation should be penalized more than overestimation in the loss function. We analyze data from the 2004-2005 Los Angeles County homeless study using an augmentation of L1 stochastic gradient boosting that can weight overestimates and underestimates asymmetrically. We discuss our choice to utilize stochastic gradient boosting over other function estimation procedures. In-sample tted and out-of-sample imputed values, as well as relationships between the response and predictors, are analyzed for various cost functions. We gratefully acknowledge numerous helpful discussions with Greg Ridgeway. The work of Brian Kriegler and Richard Berk was supported by the National Science Foundation under grant SES-0437179, \Ensemble Methods for Data Analysis in the Behavior, Social, and Economic Sciences."
Subfields: Epidemiology, General Health Professions, Sociology and Political Science
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Epidemiology
In General Health Professions
In Sociology and Political Science
- Lawrence D. Brown
- Tommaso Cai
- Anru R. Zhang
- Jamie H. Morgenstern
- Judy Hoffman
- Richard A. Berk
- Cynthia Rudin
- Kush R. Varshney
Citations in the corpus
- Semi-supervised inference: General theory and estimation of means (2016, methodology): “These predictors have been used and were known to have a useful correlation with the response [Kriegler and Berk (2010)].”
- Asymmetric Loss Functions for Forecasting in Criminal Justice Settings (2011, background): “Quantile stochastic gradient boosting may well turn out to be the best choice (Kriegler and Berk 2010), but a discussion is beyond the scope of this paper.17 To date, there seems to be no principled way to build in asymmetric costs for other than linear loss functions.”
- Predictive Inequity in Object Detection (2019, background): “Examples of the latter include predictive policing, recidivism prediction, and the allocation of social welfare resources (Rubin, 2010; Maloof, 1999; Kriegler & Berk, 2010; Chouldechova et al., 2018).”