0

I used nls package to analyze non linear model (power curve, y= ax^b).

cal<- nls(formula= agw~a*area^b, data=calibration_6, start=list(a=1, b=1))
summary(cal)

enter image description here

enter image description here

What I want now is to force intercept (a) to zero to check something. In Excel, I can't set intercept for power curve. In R, is that possible to set intercept?

If possible, could you tell me how to do it?

enter image description here

Jin.w.Kim
  • 599
  • 1
  • 4
  • 15
  • 1
    Just remove `a` from the formula (and the list of starting values) – Hong Ooi Oct 11 '20 at 00:47
  • 2
    It appears `nls` honors the same convention as for formula and therefore `nls(formula = agw ~ a * area^b - 1, data = calibration_6, start = list(a = 1, b = 1))` should do it. – Chuck P Oct 11 '20 at 00:48
  • Thank you for your answer. If I do according to HongOoi, it's y= x^1.32. And if I do according to @ChuckP , it's y= 0.96x^1.35.To be frankly, I don't know the definition what forcing intercept to zero exactly. Which equation is correct? – Jin.w.Kim Oct 11 '20 at 03:29
  • Thank you for your answer. If I do according to @HongOoi , it's y= x^1.32. And if I do according to ChuckP , it's y= 0.96x^1.35.To be frankly, I don't know the definition what forcing intercept to zero exactly. Which equation is correct? – Jin.w.Kim Oct 11 '20 at 03:29
  • Dear. @G.Grothendieck I simply want to know if my graph is starting from zero, how the equation is changed. Like linear model, we can set intercept as zero and see the equation. Like this, I want to know how my power curve equation is changed if the graph starts from zero. – Jin.w.Kim Oct 11 '20 at 16:37

1 Answers1

0

y~ x^b model is what I considered first,

 nls(formula= agw ~ area^b, data=calibration_6, start=list(b=1))

but, I also found other way, please check the below link.

How to calculate R-squared in nls package (non-linear model) in R?

Jin.w.Kim
  • 599
  • 1
  • 4
  • 15