1

There are lots of long texts about Sensitivity Analysis but as a programmer I get easily bored or can't understand it.

Can someone briefly explain Sensitivity Analysis to a programmer?

Thanks.

1 Answers1

2

Sensitivity analysis covers a lot of different things. Then, since you are a programmer, let me suppose that you make a call to a function $F(a,b,x)$ which returns to your program the profit $P_0$ of a plant running at operating conditions $a$ and $b$. Now, change your first set point to $(a+\Delta a)$ and $b$ al call $F(a+\Delta a,b,x)$ to get $P_1$. This gives you the sensitivity of the profit to the first set point value. This looks like derivatives, isn't ?

http://en.wikipedia.org/wiki/Sensitivity_analysis is a good and simple view of it and I just quote the beginning of what they write

Sensitivity analysis is the study of how the uncertainty in the output of a mathematical model or system (numerical or otherwise) can be apportioned to different sources of uncertainty in its inputs.

  • Thanks. So what Sensitivity analyse brings to us? Does it bring a mathematically easy solution for finding which condition changes profit more? Or it is just a way to explain the purpose of the analysis? I can use computer to see which condition changes output more, but it will take lots of time. – ramazan polat Mar 17 '14 at 13:52
  • 1
    @RamazanPolat In a past life I did a lot of fitting of multi-parameter models to large sets of data. The models and fits are never perfect, so one of the FIRST thing you do is a sensitivity and correlation analysis. If a parameter has low sensitivity, you might choose then to fix it to a nominal value rather than allow it to float, because loose yet useless floated parameters can induce correlations. These will usually show up in the correlation analysis as well. – uhoh Dec 23 '20 at 19:48
  • @RamazanPolat It's not a process that's easy to explain in a comment, so if you are still interested, ask further in Statistics SE – uhoh Dec 23 '20 at 19:48