Skip to content

CHRO data with model MQ

This analysis dataset attaches HCT company-year match quality to the reusable public-company CHRO and top-HR panel.

flowchart LR
    chro[(Public-company CHRO data)]
    mq[(Company-year MQ)]
    join[Collapse MQ to gvkey-year<br/>and join]
    output[(CHRO data with model MQ)]

    chro --> join
    mq --> join
    join --> output

    classDef source fill:#e8eef7,stroke:#52739e
    classDef process fill:#fff0cc,stroke:#a66b00
    classDef data fill:#eee5f5,stroke:#76528c
    class chro,mq source
    class join process
    class output data
Inputs
Dataset Location
Public-company CHRO panel /data/full/raw/chro/revelio_chro_public_firm_year.parquet
Full-training company-year MQ /data/full/analyses/full_training/company_year_mq.parquet
Build
Component Location
Builder shared/data/build_datasets/join_chro_company_year_mq.py
Runtime hct-tower-model-cuda-… (CPU execution)
Entry point python -m shared.data.build_datasets.join_chro_company_year_mq
Output
Dataset File / location Grain Key and variables
CHRO data with model MQ revelio_chro_public_with_model_mq.parquet in /data/full/raw/chro/ Public firm-year Key: gvkey, year
Variables include company_id, hr_exec_score, n_scores, model_mq

When several Revelio firms map to one gvkey-year, the builder computes MQ weighted by n_scores. It writes the result as model_mq, preserving hr_exec_score as the separate title-ranking measure.

python -m shared.data.build_datasets.join_chro_company_year_mq \
  --mq-path /data/full/analyses/full_training/company_year_mq.parquet \
  --min-n-scores 10 --year-start 1985

There is currently no project Slurm wrapper for this builder. Run the direct command in a CPU job using the model CUDA image.

The reusable source panel is documented under HR leadership data. This joined output belongs to the analysis layer because it depends on HCT scores, despite its legacy /data/full/raw/chro/ path.