Methods, code and data

Work through the model

USAJOBS figure and benchmark results · Figure PDF for the SI

Three executed notebooks use the full assembled USAJOBS corpus. Download the package for the aggregate inputs, O*NET data, benchmark outcomes and code.

  1. JAAT-coded postings → occupation-task profiles
  2. Profiles → RCA and relatedness
  3. Relatedness → Figure 2 using USAJOBS

Download all three notebooks, code and data (15 MB) · Read the run instructions

Notebooks 2–3 run with included aggregate data. Notebook 1 downloads public JAAT-coded postings and metadata; see the run instructions for inputs and dependencies. These notebooks start from saved JAAT codes, not raw-text inference. USAJOBS uses all supplied postings with weight 1. The replication applies the existing occupation-task review exclusions, as described in the notebook package; this is not a separate review of USAJOBS links. It is a worked replication with another corpus, not the paper’s NLx estimates.

The paper’s analysis

Sampling, review, population counts and benchmark audit · Paper code, released scores and reproduction · Benchmark calculations in Python or Stata

Excluded task–occupation links (CSV) · Paper figure: workbook and code

Extensions and source data

Incident-linked task explorer · Task lists, evidence and code · CFOI and SOII sources and methods

Agency profiles · JAAT on GitHub · JAAT.app

How AI is used

Use the linked code, source versions and review records to inspect each step. Some construction inputs require authorized access.

Website layout and explanatory text were prepared with generative AI (Codex). Data sources, model methods and code.

The National Labor Exchange (NLx) Data Trust bears no responsibility for the analyses or interpretations of the data presented here. The opinions expressed herein, including any implications for policy, are those of the authors and not of the NLx Data Trust members or others.