Online coding and Python classes in West Bridgford

From the Food Standards Agency's open Food Hygiene Rating data for Rushcliffe, the learner keeps the 192 establishments whose coordinates fall in a rectangle over West Bridgford, and uses only their address lines. From OpenStreetMap comes the list of 454 named streets in the same rectangle. The job is to find, for every address, which street it is on. Stage one asks whether any line of the address is exactly a street name. Stage two strips house and unit numbers, lower-cases, removes punctuation and expands abbreviations, then tries again. Stage three hands what is left to Python's difflib.SequenceMatcher, which scores two strings from 0 to 1 by their longest shared runs of characters, and accepts a match above a chosen threshold.

Cleaning did most of the work: 187 of 192 addresses were matched before any fuzzy scoring. The three good fuzzy matches were all the same kind of thing, a one-letter difference in how a name was spelled between the two sources. Loosen the threshold and errors arrive at once: at 0.75 a phrase containing "Trent Bridge" is matched to an unrelated street with a similar-looking name, and at 0.6 the town's own name is matched to a road. To test matches independently, the program checks whether the chosen street actually passes within 150 m of the address's coordinates; that confirmed 47 of the 52 exact matches and 132 of the 135 cleaned ones, the misses being down to approximate coordinates. A similarity score says two strings look alike. It does not say they mean the same place.

That habit of logging and checking is what West Bridgford learners take from the project into every later piece of Python, with or without an AI helping. The longer argument is in why teenagers should still learn to code in 2026.