How the forecast works
Short version: what a car park has actually done on mornings like this one is the best guide to what it will do today. We have the record, and we show you the mornings behind every answer, so you can judge it for yourself.
What it actually does
It remembers
A live occupancy number tells you about right now and nothing else. We have the record of what every car park actually does, morning after morning, and that is what makes tomorrow answerable at all.
It compares like with like
Today is measured against the days most like it at this car park. A Tuesday in term time is not a Saturday, and a public holiday is neither.
It shows its working
Every forecast names the real days behind it and what time each of them actually ran out. You are never asked to take a number on faith.
It admits when it does not know
When those days disagree with each other, we say so rather than showing a confident-looking time. Some car parks, and some kinds of day, simply do not have enough history yet.
Two different kinds of full
We always show both, and never merge them, because the gap between them can be hours. On 7 August, Ashfield passed 95% at 8:30am and did not actually run out until 11:50am. Calling the first one "full" would have been three hours early.
| No spaces | Every bay taken. A fact, and the one we lead with, because it is checkable. |
| Hard to find a space | Over 95% full. A judgement, and the more cautious thing to leave by: the official capacity counts disabled bays most drivers cannot use. |
How accurate it is
We re-ran the forecast over every morning in our history, using only what was knowable at the time. 11,036 forecasts, all 39 car parks, scored against what actually happened.
“You’ll get a space.”
The other 5, it had gone before you arrived.
Right 95 times in 100
“It runs out around 7:45.”
The other 16, it stayed quiet and you needn’t have rushed.
Right 84 times in 100
The exact minute is the hard part. When a car park does run out, we are usually within about a quarter of an hour of the time we gave you. Close enough to decide whether to leave now, not close enough to set your watch by. That is why every forecast also says how much the days behind it agreed, and why we say so plainly when they do not.
We would rather warn you too early than too late
Our two kinds of mistake do not cost you the same thing.
We warn you and we are wrong
You left a bit earlier than you needed to, or made other arrangements: a lift, a bus, driving the whole way. Mildly annoying. You still got to work.
We tell you it is fine and we are wrong
You drove to the station, found nothing, and now you are circling a full car park watching your train leave. Twenty-five minutes gone and no good options left.
Only the second one really hurts, so we deliberately lean towards warning you. It costs us a little on the headline numbers and we think that is the right trade.
What it will not tell you
· Anything after about 2pm. Evening event parking behaves nothing like a commute, and this morning’s shape says nothing about it.
· Much about public holidays. There are only about eight a year, so there is very little to compare against, and we say when a pool is too thin rather than guessing.
· What a brand-new car park will do. It needs history before it can have a pattern.