fedorovvvv/oh-my-measure-brand-size-charts
oh-my-measure brand size charts 496 size charts published by 132 clothing brands (EU, FR, INT, IT, JP, RU, UK, US), transcribed from each brand's own size guide and normalised into a single long-format table of 9132 rows. Read this first: rights No licence is granted over these numbers, and none is claimed. The numbers are facts about goods a brand sells — what chest girth size 40 means in that shop. They are not our work, so we do not license them from our own… See the full description on the dataset page: https://huggingface.co/datasets/fedorovvvv/oh-my-measure-brand-size-charts.
oh-my-measure brand size charts
496 size charts published by 132 clothing brands (EU, FR, INT, IT, JP, RU, UK, US), transcribed from each brand's own size guide and normalised into a single long-format table of 9132 rows.
Read this first: rights
No licence is granted over these numbers, and none is claimed.
The numbers are facts about goods a brand sells — what chest girth size 40 means in that shop. They are not our work, so we do not license them from our own name: not CC BY-SA, not CC0, not anything. Granting a licence means disposing of a right, and the rights here are not ours.
This is exactly why this dataset is published separately from `oh-my-measure-size-grids`, which does travel under CC BY-SA 4.0. Mixing them would extend that statement over material it cannot cover.
What is ours — the selection, the cross-checking, the normalisation and the shape of this export — we also claim nothing over. Take it.
Brand names are used nominatively, to say whose chart it is. If you hold rights and think a chart should not be here, say so and it will be removed — every row carries grid_id, so withdrawing one brand is deleting its rows and rebuilding.
Full statement: RIGHTS.txt.
Why long format
496 charts are heterogeneous by column: one brand prints bust, waist and hip, another prints bust only. A single wide table would pad empty columns down the full height; one file per chart would mean 992 files.
In long form a row is (chart, label, measure) and a missing measure simply produces no row — rather than an empty cell a consumer has to tell apart from zero. Pivot it back if you want width:
import pandas as pd
df = pd.read_csv("brand-grids.csv")
wide = df.pivot_table(index=["grid_id", "label"], columns="measure",
values=["low_cm", "high_cm"])Files
Join brand-grids.csv → passports.json on grid_id, and → brands.csv on brand. Provenance sits apart because repeating a source paragraph on each of 9132 rows would serve nobody.
Units
Girths in centimetres — that is how the sources print them. Column names say so (low_cm, high_cm).
Provenance
Every chart names its source URL and its capture date. All 504 sources carry both.
Not included
Brands with no transcribed chart — the app's catalogue is longer, but a claim about numbers is more expensive to earn than a claim about existence. Bras, footwear and accessories are out of scope: different measures.
Source
Produced by oh-my-measure. Live copy: `/dataset-brands/`.