---
license: cc-by-4.0
language:
  - pl
  - en
pretty_name: Poland Rental Market Medians
tags:
  - poland
  - housing
  - rent
  - real-estate
  - economics
  - cost-of-living
size_categories:
  - n<1K
---

# Poland Rental Market Medians

Median asking rents for residential rentals across Poland, by city, district and
flat size, measured from live listings. The collection pipeline runs nightly;
these downloadable editions are **periodic**, not automatically refreshed nightly.
Use the dated live API for the current serving snapshot.

- **Snapshot:** 2026-09-13
- **Coverage:** 49 cities, 96 city districts, 1–4 room flats
- **Sum of city-level sample sizes (district rows overlap these):** 22,036
- **History:** monthly series from 2026-06 to 2026-09
- **Licence:** CC-BY-4.0 — free to use commercially, attribution required
- **DOI:** [10.5281/zenodo.21892638](https://doi.org/10.5281/zenodo.21892638) — permanent, resolves to the latest archived edition (which may lag the API)
- **Source:** <https://polandrent.com>

## Where else this lives

| | |
|---|---|
| Canonical repository | [GitHub](https://github.com/thomasthinks/poland-rent-data) |
| Archived with a DOI | [Zenodo](https://doi.org/10.5281/zenodo.21892638) |
| ML ecosystem mirror | [Hugging Face](https://huggingface.co/datasets/polandrent/poland-rent-medians) |
| Data-science mirror | [Kaggle](https://www.kaggle.com/datasets/polandrent/poland-rental-market-medians) |
| Entity record | [Wikidata](https://www.wikidata.org/wiki/Q141003542) |

## Files

| File | Rows | What it is |
|---|---|---|
| `poland-rent-medians.csv` | 470 | Current snapshot. One row per city × district × room-count. |
| `poland-rent-medians.json` | — | Same data plus metadata, as served by the public API. |
| `poland-rent-history.csv` | 621 | Monthly city-level series. Forward-only. |

## Columns

| Column | Meaning |
|---|---|
| `city` | City slug (ASCII, e.g. `warszawa`, `lodz`). |
| `district` | District slug, empty for city-wide rows. |
| `level` | `city` or `district`. Lets you filter without knowing the sentinel. |
| `rooms` | Number of rooms (1–4). Polish listings count rooms, not bedrooms. |
| `median_pln` | Median monthly asking rent in PLN. |
| `sigma` | Fitted lognormal dispersion. Higher means a wider local spread. |
| `n_listings` | Listings behind that median. Treat small `n` with caution. |
| `snapshot_date` | Date the snapshot was computed (UTC). |

## Method

Listings are scraped nightly from three Polish rental portals (Otodom, OLX,
Morizon) across all 16 voivodeships, deduplicated across sources by a two-stage
entity resolution (exact source-id, then cross-source fuzzy block-and-score),
and aggregated over a 14-day active window. Medians are computed with a
lognormal fit and outlier trimming. Districts are used where the sample supports
it and otherwise fold into the city aggregate.

**These are asking rents, not signed rents.** They measure what landlords list,
and signed rents may differ. Fees and flat quality vary across listings. `n_listings` is published on
every row precisely so you can judge how much weight a given cell carries.

**The history is forward-only.** Listings disappear when they let, so a month
that was not captured cannot be reconstructed afterwards. The monthly series
starts when collection started and grows from there.

## Correction and provenance

The August 11, 2026 edition contained 528 cells, including 60 district cells
whose median, sigma and n exactly matched the old modelled seed. Those values
were not verified listing observations. The site emitter removed the seed
overlay on September 4; the mirrors missed that correction until this edition.
Use editions dated September 13, 2026 or later for measured-only current cells.
Older Git/HF/Kaggle versions and archived DOI editions remain historical records;
their files are not rewritten. Their current-grid sample counts must not be
interpreted as entirely measured. The monthly city-history file is separate.

The current month in history is partial and uses the latest captured snapshot,
not a final month-end value. City coverage can change. Sum only city-level samples
when counting observations across the current grid: district samples overlap.
A provenance manifest records the exact source commit and file SHA-256 digests.

## Citation

If you use this downloadable edition, cite its snapshot date and the repository,
and retain `provenance.json` with the source commit and file hashes. See
`CITATION.cff`, or:

> PolandRent (2026). *Poland Rental Market Medians*, snapshot 2026-09-13 [Data set]. https://github.com/thomasthinks/poland-rent-data

The [Zenodo concept DOI](https://doi.org/10.5281/zenodo.21892638) identifies archived editions, which may lag
this download. Only cite a Zenodo version DOI for these files after verifying
that its snapshot and file hashes match; an export date is not proof of archival.

BibTeX:

```bibtex
@dataset{polandrent_2026_rentmedians,
  author    = {{PolandRent}},
  title     = {Poland Rental Market Medians},
  year      = {2026},
  version   = {2026-09-13},
  publisher = {PolandRent},
  url       = {https://github.com/thomasthinks/poland-rent-data}
}
```

## Updates

Published periodically from the nightly collection pipeline. Check the snapshot
date in every row and the provenance manifest; mirrors can serve different editions.
Use the [live API](https://polandrent.com/api/v1) when you need the current serving snapshot.