Support for the 2026-06-26 release and its new monthly-aggregate schema, plus a live fallback so releases published after a package version was built remain reachable.
release_2026_06_26 (April and May
2026 data, monthly cadence, Artifacts metrics, "Cadences" report).aei_index() recognises the monthly filename convention
(aei_<source>_<date>.csv; the raw/enriched variant was dropped
from filenames) and the new monthly schema. The variant argument
is ignored for monthly releases. If a release breaks the filename
convention expected for its epoch, the other epoch's pattern is
tried before erroring.aei_geography() understands the monthly schema's geo_level
column and ISO 3166-2 subregion codes, gains a "subregion" level,
maps "state_us" onto US subregions (geo_id starting "US-")
for monthly releases, and now defaults to release = "latest".aei_compare() chooses its default join keys from the schema the
two releases share (by = NULL is the new default) and filters
monthly releases, which can ship several calendar months in one
file, to their most recent date_start window before joining."latest", ids, and dates) now falls back to
the live Hugging Face directory listing when an identifier is not in
the bundled table, so releases published after this package version
was built resolve instead of erroring. The listing is fetched at
most once per session; offline use degrades to the bundled list.aei_index() reports the file size before starting a download
larger than 50 MB (the monthly usage CSVs exceed 200 MB).aei_labor_market() fetches the standalone labor_market_impacts
tables ("job_exposure", "task_penetration") that Anthropic
publishes alongside the dated releases.aei_concentration() documentation claimed the input scale was
detected automatically; it is not, and the docs now describe the
actual behaviour. A new rescale argument normalises shares to
percentages summing to 100 before computing HHI and CR_n, for use
on filtered subsets whose shares no longer total 100.aei_cite() and aei_releases() now carry a report URL for the
2026-03-24 release (previously NA).CRAN maintenance release addressing a check-system issue raised by the CRAN team on 2026-05-31.
\donttest{} example for aei_link() fetched a release file into
the persistent user cache (tools::R_user_dir()) rather than a
temporary directory, so it left data behind on the CRAN check
machines. The example now redirects the cache to tempdir(), matching
every other example in the package.aei_cache_dir() now redirects to a session-temporary directory when
it detects an R CMD check run, so the package cannot write to
persistent storage during checks even if an example, test, or vignette
omits the redirect.bls.gov/soc/ reference link from the help page and
README. The page returns HTTP 403 to automated URL checkers; the
Standard Occupational Classification is still cited by name.Initial release.
aei_releases() lists available Anthropic Economic Index releases on
Hugging Face, blending bundled metadata with a live listing.aei_files() returns the recursive file tree of a release.aei_index() fetches the canonical usage CSV for a release, handling
the wide-to-long schema change between the 2025-03-27 and 2025-09-15
releases automatically.aei_download() is the lower-level escape hatch for fetching any
file from a release by path.aei_clusters() returns the request-hierarchy tree (Clio output) as
a parsed nested list. Released from 2025-09-15 onwards.aei_tasks() returns the bundled O*NET task statements (released
with the 2025-03-27 snapshot).aei_geography() filters the long-format enriched table to country
or US-state rows. Released from 2025-09-15 onwards.aei_compare() produces a release-on-release diff of the same metric
across two AEI snapshots, returning shared keys plus value_a,
value_b, delta, and pct_change.aei_link() is a generic merge helper that preserves the aei_tbl
class and provenance metadata. Use it to splice the AEI to your own
data on a shared key (country code, occupational identifier, cluster
name, etc.). Supports left, inner, and full joins.aei_concentration() computes the Herfindahl-Hirschman Index, top-N
concentration ratios (CR4 by default), and Shannon entropy on a
vector of usage shares. Auto-detects the share column and supports
per-group computation.aei_cite() returns a citation in plain text, BibTeX, or bibentry
form, including the methodological source paper of Handa et al.
(2025) by default.aei_cache_info(), aei_cache_clear(), and aei_cache_dir() manage
the local data cache.aei_tbl class with print, summary, and [ methods that
preserve the aei_query provenance attribute (release id, source
URL, fetch timestamp) across subsetting.