## How “Retrospective Prediction” works

I have posted earlier about Climate Science being reduced to “Retrospective Predictions”.

This is how it works:

I missed a flight today. I had predicted that I would arrive  30 minutes before check-in closed. But I was wrong. I arrived 37 minutes after check-in had closed. My prediction model was just not good enough.

My prediction model assumed a certain average velocity for my car, an empirically determined period for parking the car and getting to the check-in desk, an allowance of 15 minutes to stop for coffee along the way and an allowance of 10 minutes for inaccuracy of calculation. This gave me my starting time which – in the event – led to my being 37 minutes late for check-in.

I now applied “Retrospection” to my “Prediction”. Effectively this meant choosing which of my assumptions was wrong and where I would apply a “fudge factor”. I realised that I had not accounted for road works along the way. I therefore added in a period for “delays due to road-works” such that my total transit time was increased by precisely 67 minutes.

I then redid my calculation. Lo and Behold! My “Retrospective Prediction” was now spot on. It confirmed for me that I had been late for check-in.

(My addition of time for “delays due to road works” is a very simple but powerful factor and is given by the following equation:

delays due to road works (minutes) = 0.5 x number of days elapsed from 1st January to day of travel

In the present case, today being 14th of May it is the 134th day of the year and it is obvious that

delays due to road works = 0.5 x 134 = 67 minutes).

I am travelling tomorrow. So I shall be testing my new “Retrospective Prediction” in retrospect the day-after-tomorrow.

Tomorrow I will also try not to use the wrong departure time.