regression line, residuals, linearity


 To start, click here:
 
  • Regression line

  • Residuals

  • Linearity


Residuals are the difference between the regression line ("model") and the experimental points. 

If the hypothesis of linearity (justifying the linear regression) was correct, then the residuals are completely random

Has significance only if the "linear" hypothesis was good,  see following images...

So, let us be carefull with the correlation coefficient...



Excel tools for linear regression :
  • The most simple: The trendline ,but there miss informations, in particular the standard deviation (and so uncertainty) of the slope...
  • Less convivial but more completeLINEST
(in french: DROITEREG)