Schools in India

What “covers grades” means

A school matches a request for grades LO–HI when its own range starts at or below LO and ends at or above HI. The choice matters: filtering for grades 8–10 returns 271,355 schools under this rule, 752,643 if a school need only offer any of those grades, and 16,292 if its range must be exactly 8–10.

Class range unrecorded

635 schools carry no class range in the source and are shown as Not recorded. They are left exactly as found rather than guessed at, and are excluded whenever you filter by grade. 365 of the 635 are in Tamil Nadu — 57% of the anomalies from 3.6% of the records — which reads as a state reporting convention rather than a real population of pre-primary institutions. We do not describe them as pre-primary, because the data does not support that.

One correction, applied

15 records recorded their class range backwards — a start higher than the end. Under the covering rule a reversed range matches nothing, so these schools were invisible in every grade filter without ever raising an error. The two values have been interchanged. All 15 are in Rajasthan; a defect confined to exactly one state out of 41 is not chance. The correction is an inference and has not been confirmed against UDISE directly. The uncorrected records are retained, so the original extract remains reproducible.

Schools filed under an organisation

2,255 schools are filed under an administering body instead of a state: Kendriya Vidyalaya Sanghathan (1,324), Navodaya Vidyalaya Samiti (664), MSRVVP (118), IAF EC Society (113) and Navy Education Society (36). The source records no geographic state for them anywhere, so a state filter will not find them. They appear in the State list under their own names.

About the CSV download

The download returns exactly the rows your filters select, as UTF-8 CSV, with three columns more than the table shows: school_id, the source’s own record identifier, plus igod_state_name and igod_district_name — the same places matched to the IGOD directory’s canonical spellings, so SOUTH TWENTY FOUR PARGANAS also appears as South 24 Parganas. The UDISE columns are kept as the source writes them; use whichever suits your join. Names are reproduced as the source holds them, and the source is untidy: 2,918 school names contain a line break, 35 contain a double quote and 24,131 have leading or trailing spaces. These are properly quoted and escaped, so any real CSV reader — Excel, pandas, R — will load the file correctly. Counting lines with a text editor or wc -l will overcount, because a quoted name can span two lines.

There are two download paths, and they differ only in row order. A filtered download is assembled here in your browser, one district at a time, and is ordered by school_id within each district rather than globally. The whole directory is a pre-built gzipped file — 29 MB compressed, 125 MB expanded — served in a single request and ordered by school_id throughout. The columns are identical either way.

A source column marks where each record came from. 16,95,443 are UDISE records. A further 1,045 — 779 Cambridge and 266 IB — are international schools that hold no UDISE code, taken from those boards’ own school finders and carrying the identifiers CAMBIND-0001 and IBIND-0001 onwards. Those directories publish no grade range or pincode, so both are blank for these rows — which also means they never match a grade filter.

Some of these overlap with UDISE records. Only 656 of the 1,045 are genuinely absent from UDISE. Another 322 match exactly one UDISE school in the same state, and 67 match several. Both rows are kept — the international listing records an accreditation that UDISE does not — and the overlap is named rather than hidden: duplicate_status and duplicate_of_udise_code let you collapse them. Matching is exact on the name once punctuation is stripped, within a state, so these counts are a floor. If you are counting schools rather than looking them up, exclude rows where duplicate_status is not ‘unique’.