PR Data Observatory · Foundation for Puerto Rico

Data management

The account of every data file the observatory publishes: what it holds, how many records, fields and values it has, its size, when it was first published and last updated, which dictionary items document it and whether it needs a compliance review. The figures are counted from the files at each build; the definitions are at the end of the page.

This page is generated automatically from the observatory's repository; only the Foundation for Puerto Rico team can change the data, through reviewed changes.

Published data files

Find a file Kind Compliance review

Records, fields and values are counted for data files only: documentation (the dictionary, the catalog, the build information, the wish list) and the place-name index are listed and sized, not counted. The table does not list registry.json, the file this page reads. Sizes are bytes as published (1 MB = 1,048,576 bytes).

Growth

What is not published

Some data are counted but never drawn or named, and some are not published at all, to protect people's homes and privacy, sensitive services and contact details. They appear here as counts only: no names and no locations. Each count is read from the published file named beside it at each build.

Definitions

The terms of the table, as the build counts them (ledger.py). The counts cover the published files of the data folder and its geo subfolder.

Record: An observation unit. In a GeoJSON file, a feature. In a table, a row: a place, a stop, a trip, a road segment, a census block. In a series file, an area in a period (a municipio in a year, a barrio in a five-year window) that holds at least one value. A file with several tables (patterns, routes and stops of each day) adds the rows of all of them.
Field: A variable the records carry: a GeoJSON property, a series variable, a column. A margin of error is a field of its own. A cell that holds a list (24 hourly counts, the stop times of a trip) is one field. A field is counted when at least one record has a value in it; in a file with several tables, the fields of each table are counted.
Value (data point): A non-null observation: one field of one record that holds a value. A list cell counts its elements. Null, absent and empty text are missing and do not count. Not values: metadata (titles, labels, notes, sources, build dates, methods and checks), geometry coordinates, the area ids and years a file stores once as a lookup, and lookup tables (a legend, a price index, a search index).
Vertex: A position of a geometry: a coordinate pair of a polygon or line (the closing position of each ring counts), the latitude and longitude of a point, a point of a route path, a node or shape point of the road network. Vertices are reported apart from values.
Size: The bytes of the file as published, with line ends counted as one byte; 1 MB = 1,048,576 bytes. The limit of a GitHub Pages site is taken as 1,000 MB.
First published: The date a file first reached the published site. It never moves.
Last updated: The date its content last changed: the content hash of the file (SHA-256 of its JSON in canonical form, without the build dates written inside it) differs from the one before. A rebuild that writes the same data on another day changes neither date. Dates up to 2026-10-19 come from the history of the repository (the day each change was merged into the published branch); from then on, the date of the build that changed the content. Dates are Puerto Rico dates.
Documented by: The number of data dictionary items that describe fields of the file. The link opens them in the dictionary.
Compliance review: The state of the reviews recorded in the data contract. Open: a review of personal data, of a license or of compliance has not been completed, with its reason. Cleared: every review was completed, on the date given. License not stated: no review is open, and at least one of the file's sources states no license; a file takes the most restrictive license class of its sources, and the cell names the source that sets it. Re-use of that source's data must be confirmed with its publisher. Not required: nothing to review. Flags: Federal grant content (federal grant money or a grant-funded program: Finance and Compliance review comes before external communication), Counts only (sensitive items that are counted but not drawn or named), License not stated (at least one source states no license, named in the cell), Licensed, not open (a source licensed to FPR and not open, named in the cell).
Totals: Records, fields and values add the data files; size adds every file. A field or a value that two files repeat counts in each, so the totals measure what is published, not how many distinct facts it holds. The totals row of the table adds the files shown.