1
$\begingroup$

Could somebody kindly explain how you can express something as a covariance?

A paper I am reading contains this term:

$$\sum_in_i(p_i-p)(b_a+\sum_t\sum_jr^t\frac {n_j} {n_i}b_c)$$

The author then says that the term "can be expressed using a covariance" as follows:

$$N\mathbb C[p_i,(b_a+\sum_tr^t\sum_j\frac {n_j} {n_i} b_c)] $$

N is the sum of all n and $p$ is the average of all $p_i$.

It's a fascinating paper, but I've hit a block based on my neophyte maths. Please could somebody put me out of my misery and explain how such a covariance can be found?

Would love to hear an explanation so I'll be able to understand in the future!

$\endgroup$
2
$\begingroup$

Covariance of two random variables $X$ and $Y$ is defined as:

$$ \mathbb C[X,Y] = E_{x,y}[XY] - E_x[X]E_y[Y] $$

where $E_x[X]$ represents the expected value of r.v $X$.

$E[p_i]=p$

$$\begin{align}N\mathbb C[p_i,(b_a+\sum_tr^t\sum_j\frac {n_j} {n_i} b_c)] & =NE_i[p_i(b_a+\sum_tr^t\sum_j\frac {n_j} {n_i} b_c)]\\ & -NE_i[p_i]E_i[b_a+\sum_tr^t\sum_j\frac {n_j} {n_i} b_c] \\ &= N\sum_i\Big(p_i(b_a+\sum_tr^t\sum_j\frac {n_j} {n_i} b_c)\times\frac{n_i}{N}\Big)\\ &-N\times p\times \sum_i\Big(b_a+\sum_tr^t\sum_j\frac {n_j} {n_i} b_c \Big)\times \frac{n_i}{N}\\ &= \sum_i n_ip_i(b_a+\sum_tr^t\sum_j\frac {n_j} {n_i} b_c)-\sum_i n_ip(b_a+\sum_tr^t\sum_j\frac {n_j} {n_i} b_c) \end{align}$$

$\endgroup$
  • $\begingroup$ Thank you for such a clear explanation, rightskewed! Really helpful. (Since I'm new here it apparently won't let me 'upvote' your answer.) $\endgroup$ – Sprog Jul 14 '15 at 11:16

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.