Re: trendline gaussian

Benj,

In the case of the Oil production we wanted a curve that was 'bell shaped'. Have a look at teh attached sheet that shows four different Gamma Dist. All are possible and all could apply, but we wanted something like the first or second.

Now look at the values on the axes. The x-axis is generraly below 30 and the y-axis are all below 1, infact closer inspection below 0.5 (although this can change).

So the Oil data had x-values between 1850 and 2050, these need to be transformed to 0 to 30.

The Production numbers were in millions so I scaled them to give me a smaller number in the range for the Gamma Distribution of the form that I wanted.

These steps are known as transforming your data. (I suspect that you will not have to do this).

Now I sit and play with the a and b values till I get roughly the right shape and then let solver get the best fit.

However, if you have read the whole thread you will see that I stated early on that this was not teh best approach. Assuming that you are studying statistics at either 'A' level (High School/Junior College?) or degree level then you need to make sure that you are answering the correct question and using the correct methods.

If you are really stuck **go and ask your teacher lecturer to explain it again to you**. Take with you what you have done, including the print outs from the site I gave you and the workbook attached and your text book (have you got the recommended text book?). Remember, the chances are that the teacher/lecturer loves this subject (even if you don't). Asking someone about what they love is a gift. They will almost certainly be willing to spend the time with you.

You could spend hours asking us (and we will continue to try and help) but you might only spend 30 minutes with a teacher and have it all sorted out!

If the teacher will not help then please post the actual question (in full) and data and I will see what I can do to help you solve it.

Good luck,

A.