AI and programming classes in Royal Tunbridge Wells

The census reports two things for each of the 160 output areas in the Royal Tunbridge Wells built-up area: how many residents there are per square kilometre, and how many households have one, two, three or more people. By our sum the areas hold 22,238 households, and 7,280 of them, 32.7%, are one person living alone. Density varies enormously, from 121 to 23,491 residents per square kilometre, with a middle value of 5,521. The learner labels each area "dense" or "the rest" using a cut-off, then works out the one-person share in each group.

Raise the cut-off from 2,000 to 10,000 and the dense group's share climbs from 33.57% to 34.83%. The other group's share climbs too, from 27.47% to 32.22%. No household moved and none changed size. What changed is that 107 areas were relabelled. Together they hold 14,786 households, of which 4,909, or 33.20%, are one-person. That is a little below the old dense figure, so removing them lifts the dense group, and well above the old figure for the rest, so adding them lifts that group as well.

We did not find this by luck, and the page should say so. The program tried every pair of cut-offs from 1,000 to 12,000 in steps of 1,000, which is 66 pairs. Both shares rose in 47 of them and moved in opposite directions in the other 19. Moving the line from 2,000 to 3,000, for instance, raised the dense share to 34.43% and lowered the other to 27.28%. We then picked a pair that shows the effect clearly. A report that quotes only the convenient pair would be doing exactly what this lesson warns against.

Two more cautions. At the higher cut-off the dense group has only 32 areas, so its figure rests on 4,398 households. And these are counts for areas, which say nothing about why any household is the size it is.

A Tunbridge Wells teenager who can spot a redrawn category will read AI claims critically, and the surest way to learn that is to program the example. The longer argument is in why teenagers should still learn to code in 2026.