Page | 87
17.2 Modelling Uncertainty
Uncertainty modelling provides a method of determining how far out we might be when we estimate something. This can be very useful in all sorts of fields. If for example I were to take a bet on the height of the next person to walk down the street, I would want to know, first of all, what was the normal range of heights for people and even then I would not bet on the average. I would be much safer to bet that they fall between say 0.2 – 3.0 meters high so that I mconfd n h myb s‘s f ’.
nod o s blsh ‘s f ’b n d oknowboh h
g ndh ng ofh m su mn m yngo
estimate. s bu on’ of n u lly occu ng measurements (of which our heights are one). In this graph the x axis is height and the y axis describes the number of people in a given sample that might fall in a given height range. The German mathematician, Gauss, d sc b d h ‘ o m l
Figure 70: Modelling deviation graphically.
In this diagram the average height is in the centre and it can be seen that most people fall close to the average. The deviation of any observation is just how far it is from the average. For example if I am 1.80m tall and the average is 1.50m, my h gh “d on” would b 0.3m. o c h h g r the deviation from the average, the fewer people there will be. h fomul shows us ful p m , h “s nd d d on” whch s h squ oo of h sum of ll h deviations squared, divided by N-1 where N is the number of the sample. This number is used a great deal in uncertainty modelling and serves as a measure of the distribution around the average.
We usually use the Greek letter sigma ( ) for standard deviation.
Made with FlippingBook flipbook maker