What imibare is, why it exists, who is behind it, and how to use it.
Frequently asked questions
Imibare is the Kinyarwanda word for "numbers" or "statistics." It is also an open data infrastructure project that extracts, cleans, and republishes African governments' statistical data in machine-readable form, so you do not have to spend valuable time copy-pasting numbers out of a PDF. Rwanda and Kenya are the pilots; the architecture is designed to expand to other countries.
African government institutions publish genuinely useful data. The problem is finding it, downloading it, and doing anything programmatic with it. Data is buried in scanned PDFs, inconsistent Excel files, and poorly maintained portals. A Rwandan poverty survey arrives as a report; a Kenyan chapter of national accounts arrives as twenty spreadsheets whose tables change meaning between editions. imibare fixes the infrastructure. The data was always there; this just makes it accessible.
One person, for now. I publish without my real name because the work matters more than who is doing it. What is verifiable: the data, its sources, its methodology, and the date it was published. My name is not important; the data is.
No. This matters, so it is worth saying clearly. imibare is completely independent. It is not funded by, endorsed by, or affiliated with MINICT, RISA or NISR in Rwanda, with KNBS or the Central Bank of Kenya, or with any other government institution in any country it covers. The data on this site comes from those institutions (they remain the authoritative source), but the infrastructure, the decisions about what to publish and how, and the editorial independence are entirely mine.
Mostly stubbornness. The infrastructure cost is manageable while traffic stays modest. The work (writing the extraction pipelines, cleaning the data, maintaining the catalog) is done on evenings and weekends for now.
imibare has no advertisers, no government funders, and no sponsors with an interest in what gets published. The goal is to keep it that way.
It comes from official government institutions (NISR, BNR, MINECOFIN, RSSB, RRA and RSE in Rwanda; KNBS and the Central Bank of Kenya in Kenya), who are the authoritative sources. imibare extracts, cleans, and standardises it, then publishes the result with full provenance: where it came from, when it was retrieved, what changed in cleaning.
That said: imibare is not infallible. Extraction pipelines can have bugs. Source data can have errors. If you are relying on imibare data for something consequential (published research, journalism, policy), please verify against the original source. And if you find an error, report it.
Usually, but it depends on the source, and imibare is not the one granting it. imibare claims no copyright in factual data it did not create, and it does not re-licence other people's statistics.
Publishers differ. Some set out an explicit Creative Commons grant covering their statistics, many publish without stating terms either way, and a few reserve rights. Each dataset page records what imibare understands its publisher to state, and links to the original publication, which is the authority. Where a publisher states nothing, the page says so rather than assuming a grant. Where a publisher affirmatively reserves rights, imibare does not republish.
Whatever the terms, credit the original institution as the data source, and
imibare as the pipeline. Example: "Source: NISR (statistics.gov.rw), processed by imibare (imibare.org)"
The full position is on the terms page, and the documentation carries per-institution citation formats.
Three ways, depending on your workflow:
Python: pip install imibare, then
imi.load("rw.nisr.cpi.monthly") or
imi.load("ke.knbs.fuel-sales.annual").
Every ID begins with its country. Returns a pandas or polars DataFrame.
Direct download: every dataset page has a download button. CSV, JSON, and Parquet available.
REST API: documented at /docs. Returns the rows themselves as JSON, alongside the dataset's metadata.
Please open an issue with the dataset ID, the value you believe is wrong, and what you think it should be. Include the original source if you have it. Reporting it in the open means the correction is visible to everyone else relying on the same numbers, which an email cannot do. Every error report is taken seriously: the whole point of this project is that the data should be trustworthy.
Two different problems look identical from the outside, and both are worth reporting: imibare may have extracted the figure wrongly, or the source may have published it wrongly and imibare passed it through. You do not need to know which. If you would rather not use GitHub, hello@imibare.org reaches the same place.
Rwanda and Kenya are both live. The architecture is country-agnostic from the
start: every dataset ID begins with a country code (rw.,
ke.), and country-specific code lives in its own
directory, so the second country was a configuration change rather than a
rewrite. The expansion path is EAC first, then Sub-Saharan Africa.
What decides whether a country happens is funding, not enthusiasm. A country is roughly 650 to 1,000 hours of work by someone who knows its institutions well enough to own the pipeline for it, and that is contracted work under an agreement rather than volunteering. Rwanda cost a year of evenings, which is not a thing that repeats.
If you know a country's data landscape that well, or you want to fund coverage of one, let's talk.
No. But they are the questions I would ask if I landed here for the first time, so here we are.