Coding and AI classes in South East England

The learner downloads both raw station files, skips the header lines, and reads each month's average maximum and minimum temperature, taking their midpoint as the month's mean. Missing months are marked with dashes and estimated ones with a star, so the code has to cope with both; Manston has two gaps in the thirty years. Then it averages each calendar month over 1991 to 2020, the same window the Met Office uses for its standard averages. The yearly means come out identical to two decimal places, which makes the comparison clean.

October and April sit the same distance from midsummer, so on a planet with no oceans they would be about equally warm. Both stations are warmer in October, because the ground and air store some summer heat, but Manston by almost a degree more. Spring shows the other side: Oxford is warmer than Manston in April and May, while Manston is warmer from August to November. That is seasonal lag, and the sea is the store of heat that causes more of it. Interestingly, the range from coldest to warmest month is almost the same here, so the timing changes more than the size of the swing.

The key slip is comparing different periods. Oxford's record allows a mean for 1861 to 1890, and it comes out at 9.68 degrees, 1.36 below its modern figure. Set that against Manston's 1991 to 2020 figure and you would wrongly conclude that the coast is far warmer, when the gap is mostly the difference between centuries. The code enforces one window for both stations, and the learner writes a check that fails if the windows differ.

Checking that numbers are truly comparable is a lasting reason for a South East teenager to keep coding in 2026. The longer argument is in why teenagers should still learn to code in 2026.