2

I have found one breakpoint in my linear regression but the slope of two lines have the same decreasing slope, I was wondering if the change of the slope is necessary in recognition of breakpoints.

Penguin_Knight
  • 11,988
SiSi
  • 21
  • 2
    Can you explain what is meant by: i) you "found one breakpoint?" How did you find it? ii) What do you mean by "same decreasing slope?" Do you mean both slopes being negative? Or both slopes being negative and with similar magnitude? – Penguin_Knight Mar 10 '14 at 16:53
  • In line w/ @Penguin_Knight's questions, is it that there is a vertical shift at 1 point, but the slopes are otherwise identical-ish? – gung - Reinstate Monica Mar 10 '14 at 17:04
  • Actualy I analysed every point with linear regression and I chose the significant one (Rs greater than P value) as a breakpoint but the slope is negative in two segmented lines around the breakpoint. – SiSi Mar 11 '14 at 07:01

1 Answers1

2

Two lines can have the same slope But they can have different intercepts thus being parallel. A change of slope is not necessary but is sufficient. A change of intercept is not necessary bur is sufficient.

IrishStat
  • 29,661
  • Thank you. The slopes and intercepts have different values but the slope of two segmented lines are negative. – SiSi Mar 11 '14 at 07:04
  • your comment confuses me. Can you provide a table of all the slopes and intercepts and the actual data ? – IrishStat Mar 11 '14 at 12:10