`data/population_sample.csv` appears to be a 100-row placeholder, causing widespread "not in population sample" warnings

[asteier2026]

data/population_sample.csv appears to be a 100-row placeholder, causing widespread "not in population sample" warnings

Problem

data/population_sample.csv contains only 100 rows, which is far too small to cover the full
range of PUMS attribute values. When running compute_reid_risk, any attribute value not present
in the sample is silently skipped when fitting the Gaussian copula model, reducing the number of
attributes used for re-identification estimation and likely underestimating risk.

This also contradicts the README (line 52), which describes the file as:

data/population_sample.csv contains the 3M sample of US census data

Specific coverage gaps in the 100-row file

Attribute	Present	Expected
ST	36 of ~51 state codes	Missing 20 states including NY (36), TX (48), OH (39), PA (42), OR (41), TN (47), UT (49)
OCCP	43 of 493 occupation codes	—
ESR	5 of 7 values (missing codes 5 and 7)	—
DOB-Day	28 of 31 days (day 1 missing entirely)	—

Additional note

uniqueness.py line 48 references a different path (data/pums/pums_pwgtp_sample_.csv) that does
not exist in the repository, suggesting a larger weighted PUMS sample was intended but not
included.

Question

Is there a larger population sample file that should be used? If it is omitted intentionally (e.g.
due to size or licensing constraints), could you document how to obtain it and what format is
expected?