Allocating Interventions Based on Predicted Outcomes: A Case Study on Homelessness Services
Authors: Amanda Kube, Sanmay Das, Patrick J. Fowler622-629
AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Using data from the homeless system, we use a counterfactual approach to show potential for substantial benefits in terms of reducing the number of families who experience repeat episodes of homelessness by choosing optimal allocations (based on predicted outcomes) to a fixed number of beds in different types of homelessness service facilities. |
| Researcher Affiliation | Academia | Amanda Kube, Sanmay Das Computer Science & Engineering and Division of Computational and Data Sciences Washington University in St. Louis {amanda.kube,sanmay}@wustl.edu; Patrick J. Fowler Brown School and Division of Computational and Data Sciences Washington University in St. Louis pjfowler@wustl.edu |
| Pseudocode | No | The paper does not contain any pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide any statement or link for open-source code specific to the methodology described. |
| Open Datasets | No | The paper states, "Data for the project come from the homeless management information system (HMIS) of a major metropolitan area from 2007 through 2014." However, it does not provide concrete access information (link, DOI, etc.) for this administrative dataset, nor does it specify that it is publicly available. |
| Dataset Splits | No | The paper mentions "out-of-sample" predictions, indicating a split, but does not provide specific percentages, counts, or explicit methodology for training, validation, and test splits needed for reproduction. For example, it says: "It predicts (out-of-sample), in expectation, 2227 (43.72%) of households would re-enter the system, and 2193 (43.04%) actually did." |
| Hardware Specification | No | The paper mentions the use of "Gurobi optimization software" and the "R package Bayes Tree" but does not specify any hardware details like GPU models, CPU types, or memory amounts used for the experiments. |
| Software Dependencies | No | The paper mentions "Model fitting and counterfactual inference were done using the R package Bayes Tree written by the model s creators (Chipman et al. 2010)." and "We use this IP framework and Gurobi optimization software to find an optimal allocation". However, it does not provide specific version numbers for either the R package Bayes Tree or Gurobi optimization software. |
| Experiment Setup | Yes | These posterior samples consist of 1000 post-burn-in samples for each observation. ... The optimization problem Let xij be a binary variable representing whether or not household i is placed in intervention j. Then, the Integer Programming problem is given by ... We use this IP framework and Gurobi optimization software to find an optimal allocation for households who entered the system during each week. ... It is possible that the inefficiency of the original allocation is in part due to humans making decisions in the interests of equity. One way to potentially deal with fairness concerns like these is to make them explicit in the optimization. As an example, we consider what happens if we add a constraint that prevents any household from suffering too high a predicted cost from the change in allocation: j pijyij + 0.05 i |