Packaging Plastic Weights

An open dataset of plastic packaging weights for everyday consumer products. 60 measured products + 19 category averages, with sources, confidence levels, and barcodes.

"How much plastic is in my weekly shop?" — a question with no good public answer. This dataset is a start.

---

What's in here

data/
├── products.csv      60 products with measured/estimated plastic weights
└── categories.csv    19 category averages (for items not yet in products)
docs/
└── METHODOLOGY.md    How weights are measured, confidence tiers, limitations
examples/
└── quickstart.py     Look up a receipt item in 20 lines of Python

The schema, at a glance

Each row in products.csv is one product (one SKU, one size). The columns that matter most:

Column	Meaning
plastic_grams	Weight of plastic packaging in grams
confidence	measured (weighed in-house), category-avg (fallback to category), or inferred (estimated from similar products)
primary_material	HDPE, PET, LDPE, PP, PVC, mixed, etc.
source	Where the number came from — must be a real, traceable reference
barcode	UPC-A or EAN-13, checksum-validated. Optional
region	Where this SKU is sold (US, UK, EU, global)

Full schema in docs/METHODOLOGY.md.

Using the data

import csv

with open("data/products.csv") as f:
    products = list(csv.DictReader(f))

# Find all measured oat milks
oat_milks = [p for p in products
             if p["subcategory"] == "oat-milk" and p["confidence"] == "measured"]
for p in oat_milks:
    print(f"{p['name']}: {p['plastic_grams']}g {p['primary_material']}")

See examples/quickstart.py for a slightly bigger example that handles a full receipt.

What this dataset is — and is not

It is:

• A starting point for plastic packaging research, sustainability tooling, and consumer apps
• Honest about uncertainty — every row carries a confidence tier
• Free for any use under CC-BY-SA-4.0, including commercial, provided you attribute and share-alike

It is not:

• Comprehensive — 60 products is a seed, not a survey of the supermarket
• A substitute for primary measurement — inferred rows should be treated as ballpark estimates
• An LCA (life-cycle assessment) — we measure packaging, not full environmental impact
• An audited dataset — see METHODOLOGY for sources and limitations

Contributing

Got a kitchen scale? You can help. Most useful contributions: weighing real packaging for products not yet in the database, especially outside the US.

Read CONTRIBUTING.md for the submission process. The short version: open an issue with the new-product template, or open a PR adding a row.

Citation

If you use this dataset in research, please cite it. See CITATION.cff, or use:

Packaging Plastic Weights. (2026). An open dataset of consumer packaging weights. https://github.com/ikm365789/packaging-plastic-weights

License

Data and documentation: CC-BY-SA-4.0. You must attribute the source and share derivative datasets under the same licence.

A note on origin

This dataset was originally built for Loop — a receipt-tracking app that helps people see the plastic in their shopping. The dataset is published openly because plastic packaging data is a public good. Loop maintains it; anyone can use it; pull requests are welcome from any direction.