Form 990N Machine Learning Model

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

The output is not presumed to be high-impact and is not used as the principal basis for significant decisions/actions

Start date

2025-01-22 (day precision)Source: 1/22/2025

Withheld from public reporting

NoSource: a) No

Mission

Topic area

Information Technology

Operational functions

No function tag matched Derived by keyword rules; see methodology

Problem the AI is intended to solve

The goal of this model is to predict non-eligible Form 990N filers as part of a data driven approach to identify non-compliance within the population of exempt organizations (EOs) filling the Form 990N (Issue Control Number (ICN) 121004).

Expected benefits

Form 990-N filings provide limited information to the IRS, making it difficult for Tax-Exempt Government Entities (TEGE) to detect non-compliance among that population of EOs. This use case provides improved ability to predict non-compliance.

System outputs

The model output is an Excel workbook with 34 columns including Employer Identification Number (EIN) and a PREDICTION column.

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

Yes

Privacy Impact Assessment

Not reported

Authorization to Operate

Yes

Demographic features

k) None of the above

Training and evaluation data

3 years of transactional data for each entity is used with a combination of Hyper-parameter tuning as well as feature engineering to generate a Random Forest classification model. Model performance was assessed by primarily looking at recall and Receiver Operating Characteristic (ROC) curves.

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