Ulkaa Edu Rank: methodology in preparation

Why student demand is the largest pillar

In Bangladesh, where a student applies says a lot. Public universities hold admission tests with many applicants per seat, and the best-scoring students choose among a small set of universities. Where students actually choose to go is evidence that no international list captures.

So student demand carries 25% of the score. It is measured each admission cycle: applicants per seat, how many students a university takes in and how fast that is growing. Each edition reflects that year’s cycle, and an institution page will show demand over time.

Demand is compared inside each kind of institution, so a private university with a strong enrolment pull is not penalised for the national preference for public universities, and the score does not add that preference on top. An institution with no evidence of student demand is not ranked, however good its other numbers are.

How a score is calculated

Every indicator goes through the same steps: take the raw value, apply a transform where the data is skewed (a logarithm for counts such as publications), cap extreme values at the 5th and 95th percentile, scale to 0–100, and flip the scale where lower is better.

Indicators are averaged inside their pillar, and the pillars are combined with their weights. The result is the score, with one decimal. Every row keeps the raw value, year and source link behind each number.

Missing data and coverage

Public data has gaps. Missing data lowers an institution’s coverage, never its score: the weights of the indicators we do have are rescaled so the rest still add up to 100%.

Coverage is shown on every row. An institution with less than 60% of the weight covered is shown as not ranked because of insufficient public data, not as a low-ranked one.

Ranks, ties, bands and categories

We show exact positions for the top 30 only. After that we publish bands (31–40, 41–50, 51–75, 76–100 and 101+), because small score differences far down the list do not mean much.

Besides the overall list there are public and private lists, one list for each of the eight divisions and subject lists such as CSE, engineering and business. A category page appears only when it has at least five ranked institutions.

We do not publish “worst” lists. Unranked institutions never appear in a table, and movement arrows on the edition page show change from the previous edition.

Medical colleges

Medical colleges are scored with their own indicators: student demand, research output and impact, clinical capacity (hospital beds for each MBBS seat), web presence and international research links.

Where per-college admission merit is not yet public, demand uses a proxy such as the college’s age and seat count, and says so. The proxy is replaced as soon as better data exists.

Global recognition: QS and THE

The published QS and THE ranks are one input, worth about 10% of a university’s score. Only the published rank or band is used, converted to points: a better rank earns more points.

QS and THE scores, sub-scores and tables are never stored or shown here. They are the publishers’ copyright, and the Ulkaa Edu Rank does not depend on reproducing them.

What we do not do

We take no payment for a position, run no survey of our own, use no data that sits behind a login and publish no lowest-ranked lists. A university cannot buy a better row, and a university that asks us to remove its row cannot have it removed.

Sources and years

University Grants Commission (UGC) annual reports; admission circulars and results published by universities and the press; OpenAlex for research output and citations; accreditation registers (BAETE, BAC, the Pharmacy Council and the Bar Council); Webometrics; and the published QS and THE ranks. Each metric on an institution page links to its source and shows the year.

Known limitations

Demand evidence is thin for some institutions. Citations are counted, not adjusted for field. Engineering universities set a high initial screening bar before their admission test, which makes their applicants-per-seat look low, so their demand indicator is set to the top of the group. Some Webometrics data is older than the rest. These limits are why we publish bands.

Corrections and updates

If a number is wrong, tell us with a link to the source. We aim to reply within seven days. A fix to a published edition goes into the next one and is listed with it, because a published edition is frozen.

We add QS and THE editions as they are released: QS World usually in June, THE in September or October, QS Asia in November. The Ulkaa edition is published once a year; the release calendar shows the month.