Online coding and Python classes in Cumbernauld

The learner downloads OpenStreetMap data for a rectangle over Cumbernauld: 248.1 km of roads and 209.3 km of footways, paths, cycleways and steps. From it Python builds two networks. The walking network uses every way a pedestrian may use, leaving out motorways and trunk roads; the driving network keeps only public roads. The program then picks 600 random pairs of points found on both networks, at least 300 m apart, finds the shortest route between each pair with Dijkstra's algorithm, and divides it by the straight-line distance.

On foot, a typical trip here is about a third longer than the straight line; by road it is about 70% longer, and one trip in ten is at least 2.7 times the direct distance. For the median pair, the drive is about 20% longer than the walk, and in 18.8% of pairs the drive is more than one and a half times the walk. Short trips suffer most: under a kilometre, walkers go 46% further than the crow and drivers 90% further; a likely reason is that the first and last stretches of a journey rarely point the right way, and on a short trip they are most of it. Over longer distances the extra shrinks towards the straight line.

The numbers come with honest limits. The rectangle cuts routes that would leave it, one-way streets are ignored, and a path counts only if someone has mapped it. Change any of those and the numbers move, which is exactly why a learner should be able to rerun the analysis rather than quote it.

Checking how an answer was measured lets Cumbernauld teenagers use AI with confidence rather than faith, a fine reason to learn Python in 2026. The longer argument is in why teenagers should still learn to code in 2026.