The dataset viewer is not available for this subset.
Exception: SplitsNotFoundError
Message: The split names could not be parsed from the dataset config.
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 291, in _generate_tables
pa_table = paj.read_json(
io.BytesIO(batch), read_options=paj.ReadOptions(block_size=block_size)
)
File "pyarrow/_json.pyx", line 364, in pyarrow._json.read_json
File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
return check_status(status)
File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
raise convert_status(status)
pyarrow.lib.ArrowInvalid: JSON parse error: Column() changed from object to string in row 0
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
for split_generator in builder._split_generators(
~~~~~~~~~~~~~~~~~~~~~~~~~^
StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 101, in _split_generators
pa_table = next(iter(self._generate_tables(**splits[0].gen_kwargs, allow_full_read=False)))[1]
~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 306, in _generate_tables
batch = json_encode_fields_in_json_lines(original_batch, json_field_paths)
File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 111, in json_encode_fields_in_json_lines
examples = [ujson_loads(line) for line in original_batch.splitlines()]
~~~~~~~~~~~^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 20, in ujson_loads
return pd.io.json.ujson_loads(*args, **kwargs)
~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
ValueError: Expected object or value
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 68, in compute_split_names_from_streaming_response
for split in get_dataset_split_names(
~~~~~~~~~~~~~~~~~~~~~~~^
path=dataset,
^^^^^^^^^^^^^
config_name=config,
^^^^^^^^^^^^^^^^^^^
token=hf_token,
^^^^^^^^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
info = get_dataset_config_info(
path,
...<6 lines>...
**config_kwargs,
)
File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
IPEDS Database
Query U.S. postsecondary institution data from NCES IPEDS: directory, admissions, enrollment, completions, tuition, financial aid, graduation rates, staffing and finance. The database covers 1997 through the available 2025 files, with harmonized historical tables and separate source-preserving 2024 Cost survey tables.
November 9, 2026 release: 27,889,050 rows in 25 source tables, plus three views. The 1,416,376,320-byte file includes 30 verified final-revision corrections and adds 11,273 records from the two provisional 2024 Cost surveys. Existing table schemas and views are unchanged. See release provenance, the dictionary and the source/replay bundle.
Query it remotely
INSTALL httpfs; LOAD httpfs;
ATTACH 'https://huggingface.co/datasets/Nason/ipeds-database/resolve/main/ipeds.duckdb' AS ipeds (READ_ONLY);
SELECT year, COUNT(*) AS institutions FROM ipeds.hd GROUP BY 1 ORDER BY 1 DESC LIMIT 5;
R with DBI and duckdb
Install the dependencies once with install.packages(c("DBI", "duckdb")).
library(DBI)
con <- dbConnect(duckdb::duckdb())
dbExecute(con, "INSTALL httpfs")
dbExecute(con, "LOAD httpfs")
dbExecute(con, "ATTACH 'https://huggingface.co/datasets/Nason/ipeds-database/resolve/main/ipeds.duckdb' AS ipeds (READ_ONLY)")
dbGetQuery(con, "SELECT year FROM ipeds.hd LIMIT 5")
dbDisconnect(con, shutdown = TRUE)
DuckDB uses HTTP range requests; queries run on your machine and may need substantial memory or download data. A LIMIT does not guarantee a small scan.
Or install the Python datapond package or released R datapond package and connect to ipeds-db.
What changed and what was checked
- Thirty existing table/year partitions were updated from provisional to revised/final source members. Their provisional inputs first had to reproduce the published baseline. The corrections add 799 net rows and also change values in existing rows.
cost1adds 5,867 institution records and 419 columns includingyear.cost2adds 5,406 institution records and 540 columns includingyear. These are new tables, rather than incompatible replacements for older IC tuition tables.- This release checked 60 of the 496 existing table/year source partitions. The remaining 436 retain the previous publication and were not rechecked for newer revisions in this run. This is a bounded, verified update, not a claim that every historical vintage is current.
- The original 23 table schemas and three view definitions are preserved. Source identities, changes, validation results and reproducible inputs are included with the release.
Cost data: preserve the source meaning
The Cost files are provisional. year = 2024 identifies the source file/collection,
not the reporting period of every variable. COST1 includes periods from 2021-22 through
2024-25; COST2 includes average net price for 2023-24. Consult each variable's official
Cost dictionary before comparing periods or institutions.
unitid is BIGINT and joins to institution identifiers in other IPEDS tables. Other
original Cost fields are VARCHAR so codes, formatting, imputation flags and empty
strings remain exactly as published. An empty string is not converted to SQL NULL.
Cast numeric fields explicitly only after reviewing their documented codes.
SELECT unitid, year FROM ipeds.cost1 ORDER BY unitid LIMIT 5;
Avoid joining child tables without checking their grain. A table's row count is not
necessarily a count of institutions, and most historical joins need both unitid and year.
Known legacy harmonization limits
The revision corrections retain the existing schema rather than silently changing
column types. In hd, numeric close_date cannot represent 68 slash-formatted date
strings in the 2024 source and retains those as NULL. In gr, numeric line cannot
represent the source labels 29A, 12A, 18A and 11A; those labels remain NULL.
The 2023 GR revision changes their record counts, not the set of affected label categories.
Revision-table source comparisons therefore use the existing harmonization. The new
Cost tables, by contrast, are checked losslessly against the original source fields.
Sources and reproduction
Underlying data: NCES IPEDS, public domain. Original build pipeline by Paul Goldsmith-Pinkham, maintained here for Datapond. The release bundle supplies the bounded revision and Cost-extension scripts, official source archives/dictionaries, hashes and replay instructions. Use DuckDB 1.5.5 for the pinned replay. Read-only Python and R consumer checks must pass against the published file before treating a release as verified.
The previous baseline is retained in repository history. No automatic refresh is promised.
- Downloads last month
- 1,227