I am plotting the nonlinear regression on some fatigue data. My goal is to create a line that adheres to the slope line (in red). This line should sit beneath the lowest point on the left side of the graph.
I asked over on stackoverflow and was given a reasonable answer. But that answer only allows the line to pass through the lowest rightmost point. I am wondering what ways mathematically can this be achieved? I have many different plots that I need to apply this logic too.
the equation I am using is $A*N^B$ where B is a constant and N is the length of cycles and A is the mean of the y values.
What I have tried. Standard deviation to find the z value and therefore farthest value under the left side of the fitted line. I then shifted the fit line down to this value. However it did not give me consistent results.
I also obtained the Convex Hull and tried to determine which point was most to the left and down but this also didn't work.
Your line is not uniquely defined. Which of the two lines here do you prefer and how would you unambiguously specify it in the general case?
For the OP: State precisely which two points in the below data set define the line you seek, and also state precisely and unambiguously why you don't want the neighboring two points.
Which two points are "outermost"?