Linerization of non-linear models (1 Viewer)

Users Who Are Viewing This Thread (Users: 0, Guests: 1)

Hello,
I'm a student from Lithuania. I have found in one forum topic your discussion about non-linear equation convertion into a linear form.. I have some problem with that...Could you help me, please? :)
For, example, I have two non-linear models:
1. y = a [1/(1+b/(c+x))]*exp(-d*x)
2. y = a*exp[b*(1-exp(-c*x))/c - d*x]
And I want to convert it into linear equation, for fitting linear regression theory, of course, I could fit a newton or other method, but there are some problems with initial values. So, I would like to see the results of fitting to these models a linear regression and compare with iterative methods.

So, could you help me?

I think, that these two models linear form will be:
1. ln(y) = ln(a) - ln(1+b/(c+x)) - d*x + err
2. ln(y) = ln(a) + b(1-exp(-c*x))/c - d*x
lnln(y) = lnln(a) + ln(b/c)-ln(b/c)-c*x -ln(d) + ln(x)

the main problem is, how to estimate unknow a,b,c,d parameters...using for example, SAS programe package.

I would waiting from you any help.

With best regards, Ms. Dovile.
 

HallsofIvy

Science Advisor
41,626
821
?? None of your suggested "two models linear form" are linear.
 

The Physics Forums Way

We Value Quality
• Topics based on mainstream science
• Proper English grammar and spelling
We Value Civility
• Positive and compassionate attitudes
• Patience while debating
We Value Productivity
• Disciplined to remain on-topic
• Recognition of own weaknesses
• Solo and co-op problem solving
Top