The Colwyn Bay projectTwo missing heights and a thousand wrong answers
Postcodes, a terrain model, one tidy-looking filter, and what zip does with lists that no longer match.
The learner starts with all 1,066 postcodes in the LL28 and LL29 districts from Ordnance Survey's Code-Point Open, looks up the position of each one, and asks a terrain service, OpenTopoData, for the ground height at that point from the EU-DEM elevation model. The service returns the heights in the same order the positions were sent. 1,064 come back as numbers; two come back empty. Both belong to postcodes on the shore at Rhos, LL28 4EN and LL28 4PR. Our reading is that the points sit where the model has no land value, but either way the gap is real.
Now the tidy-looking line. A common first instinct is to remove missing values before doing anything else, for example heights = [h for h in heights if h is not None], and then pair the results with the postcodes using zip(postcodes, heights). Python accepts it. It produces 1,064 pairs without complaint. It is also badly wrong.
What one filtered list does to the pairing, our Python run| Measure | Result |
|---|
| Pairs produced by zip | 1,064 of 1,066 |
| Postcodes silently left out | LL29 9YP and LL29 9YW |
| Postcodes given a neighbour's height | 1,003 (from the 62nd onwards) |
| Of those, given a different number | 991 |
| Typical error (median) | 8.77 m |
| Errors over 10 m | 459 |
| Largest error | 221.3 m |
Removing the first missing height slid every later height up by one place, so from the 62nd postcode onwards each one was paired with its neighbour's height, and after the second gap, with the one two places on. The two postcodes at the end of the list simply vanished, because zip stops when the shorter list runs out. The damage then spreads into every answer built on the pairs. Ask which postcode stands highest and the broken pairs say LL29 6AW at 306.2 m; its real height is 158.3 m. The true highest point in the list, LL29 6BA, sits two places further down the sorted list, which is exactly why its height was handed over.
There are two clean fixes, and the learner writes both. Since Python 3.10, zip(postcodes, heights, strict=True) refuses to pair lists of different lengths and stops with ValueError: zip() argument 2 is shorter than argument 1, which turns a silent bug into a loud one. Better still, pair each height with its postcode before cleaning anything, in a dictionary built straight from the service's reply, so the two missing values stay attached to their own postcodes and can be dealt with deliberately.
Ages 8 to 11
Two columns of cards: take one card out of the second column and see who ends up with the wrong partner.
Ages 11 to 15
Zip two short lists of different lengths in Python and find out what disappears.
Ages 15 and up
Reproduce the postcode bug, measure the damage, then fix it with strict=True and with a dictionary.
Sources and limits
Postcodes are from Ordnance Survey Code-Point Open 2026.3.0, with Royal Mail and OS data under the Open Government Licence; positions come from postcodes.io. Heights are from OpenTopoData, serving the Copernicus EU-DEM model at 25 m resolution; they describe the ground model, not any building. The filter, the pairing and every count in the table come from our own Python run.
