Let us consider a variable X with values $\{-1,1\}$, and another one Y with values $\{-2,-1,1,2\}$.

X and Y have the following joint probability function $P(x,y)$:

enter image description here

From this table, the covariance is zero.

Does this mean that there are no linear relationships between X and Y (because covariance function should measures linear relationships)?

Then, is it possible to calculate the non linear relation between X and Y or, at least, prove that their relationship is non linear?


In this case, $y$ takes the values $\pm\left(\frac{3}{2}x-\frac{1}{2}\right)$ with equal probabilities. Since $y$ take both positive and negative values with equal probabilities, we can't say that $y$ increases or decreases with $x$. However, it is not possible to deduce such relationships easily, for a general case.

  • $\begingroup$ Thank you for your answer! I edited my post... $\endgroup$ – Mark Apr 9 at 10:37
  • $\begingroup$ Can we prove that the relationship is non linear? I mean, even without knowing this relation... $\endgroup$ – Mark Apr 10 at 18:42
  • $\begingroup$ What do you mean by non-linear: $Y\neq aX+b$ ? $\endgroup$ – Geethu Joseph Apr 11 at 3:53
  • $\begingroup$ Yes I mean this. $\endgroup$ – Mark Apr 11 at 10:20
  • $\begingroup$ Such a relation holds only if for every value of $X$, there exists a unique value of $Y$. For a general case where there is a one-to-one correspondence between $X$ and$Y$, it is easy to verify if a such linear relation holds. $\endgroup$ – Geethu Joseph Apr 11 at 10:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.