Skip to content

Worker-firm-period scores

This is the central HCT output: model scores and predicted separation hazards for observed annual worker-firm periods.

flowchart LR
    periods[(Model relationship periods)]
    checkpoint[(Trained HCT checkpoint)]
    score[Score observed periods]
    output[(Worker-firm-period scores)]

    periods --> score
    checkpoint --> score
    score --> output

    classDef source fill:#e8eef7,stroke:#52739e
    classDef process fill:#fff0cc,stroke:#a66b00
    classDef data fill:#eee5f5,stroke:#76528c
    class periods,checkpoint source
    class score process
    class output data

Grain: one scored annual worker-firm period
Location: /data/<coverage>/analyses/<mode>/worker_firm_period_scores/
Scorer: model/scoring/score_matches.py

The output preserves worker, relationship, firm, period, and split keys. It contains match_score, predicted hazard quantities, observed hazard_label, wage and job fields, and in/out-of-sample indicators. One output shard is written for each source relationship-period shard, with expected filenames and row counts recorded in _metadata.yaml.

  • training uses the checkpoint fitted on train and validation workers and preserves a meaningful out-of-sample test group.
  • full_training uses all configured splits and supplies the final MQ measure for downstream descriptive and economic analyses.
python -m scoring.score_matches --config config/config_one_percent.yaml --mode training
python -m scoring.score_matches --config config/config_one_percent.yaml --mode full_training

On Slurm, use slurm/submit.sh --task score with the same mode and config. The scores feed the disaggregated, worker-firm, company-year, and linked-executive datasets.