Public Assistance Workload Projections

Overview

Development stage

DeployedSource: c) Deployed – The use case is being actively authorized or utilized to support the functions or mission of an agency.

High-impact designation

Not high-impactSource: c) Not high-impact

Impact justification

Not reported

Start date

2018-10-01 (day precision)Source: 2018-10-01T00:00:00

Withheld from public reporting

Not reported

Mission

Topic area

Emergency Management

Operational functions

No function tag matched Derived by keyword rules; see methodology

Problem the AI is intended to solve

The use case is predicting recovery program quantities of interest using supervised learning models to include predicting the number of applicants who will apply for Public Assistance, predicting the number of PA projects that applicants will submit, predicting the number of sites that will need to be inspected per PA project, predicting the cost of delivering assistance, etc. Supervised learning models include but are not limited to the use of sample statistics, generalized linear models, decision trees, and deep neural networks for the purpose of predicting unknown quantities.

Expected benefits

1. For informational purposes: the models will produce predictions for to-be-determined quantities of interest. These quantities are often of interest to Agency personnel in the field, region, and headquarters, as well as DHS, OMB, NSC, and the White House. 2. For decisional purposes: in addition to being informative, the model’s predictions are likely to be used for decision making. Projections help inform staffing levels and timing.

System outputs

Supervised learning models produce predictions, not recommendations and not decisions (though they can be used to inform human users in making recommendations and decisions). A minimum, these supervised learning models will produce point predictions for the different quantities of interest for disaster declarations. Additionally supervised learning models may produce prediction intervals or predictive distributions as feasible and appropriate for the given prediction problem. Often these outputs will be shared via business intelligence tools (e.g., Tableau or PowerBI) for wide internal FEMA use. Some predictions may be shared to a more restricted audience through simpler means (e.g., an excel workbook) as appropriate.

Technology

AI classification

Classical / predictive MLSource: Classical/Predictive Machine Learning: Models trained on data to make predictions or classifications based on identified patterns or relatio…

System name(s)

Not reported

Custom-developed code

Yes

Public source code

Not reported

Sourcing

How it was built

Developed in-houseSource: b) Developed in-house

Vendor (as reported)

Not reported

Vendors (standardized)

None on the standard list

A vendor is the supplier named by the agency. It does not identify the underlying model or AI technology.

Data and privacy

Involves PII

No

Privacy Impact Assessment

Not reported

Authorization to Operate

No

Demographic features

Not reported

Training and evaluation data

FEMA: Historical Declaration and Public Assistance activity data U.S. Census: Housing Units Logged, Density Housing Units, Number City/Township Govs, Number of Special District Govs DHS Infrastructure: Fire Stations, Electric Substations, Dams, Ten Mile Power Lines Dept of Agriculture: Agricultural Land (sq. miles), Wetland (sq. miles), Developed Land (sq. miles)

Federal Data Catalog

Not reported

Governance

0 of 8 minimum-practice questions answered. A blank answer means the agency reported nothing; it does not mean the practice is absent.

Potential impacts and how they were identified

Not reported