# How to calculate the expected frequency of a pattern?

I'm working on a problem to find the expected frequency of a pattern. Say there is a sequence of alphabets - A, B, C and D.

I want to find the expected frequency of a pattern ACD given the sequence above. So I calculated the frequencies of A (0.4), B(0.2), C(0.1) and D(0.3) separately.

Initially I thought, multiplying the frequencies of A, C and D would suffice, i.e., 0.4 * 0.1 * 0.3 = 0.012. But, this is not what i need as I need to conserve the order of ACD.

Can anyone tell me how to proceed with this?

Thanks!!

-
The question is not at all clear. What do you mean by "the expected frequency of a pattern ACD given the sequence above"? The frequency in what? In the given sequence? Is there supposed to be some random process occurring here? – Robert Israel Aug 27 '12 at 7:09
In the sequence I mentioned "ABD'ACD'BADA", the pattern occurs only once. This I could probably say is the observed frequency. Now by learning the individual frequencies of each alphabet from the sequence, I want to know what is the expected frequency of the pattern ACD in the given sequence. – sabertooth Aug 27 '12 at 7:28
Basically, to make it short I just need know to the probability of the pattern, 'ACD' from the given frequencies of each alphabet. – sabertooth Aug 27 '12 at 7:29

If A,C,D have probabilities $4/10$, $1/10$, $3/10$ respectively of appearing in any position, independent of what appears elsewhere, then any given triple of distinct positions has probability $.4 \times .1 \times .3 = .012$ of getting ACD (in that order).
Assume that the letters $A_i$ $\ (1\leq i\leq m)$ have given a-priori probabilities $p_i\,$, and that at each position of the string one of these letters appears with the given probability and independently of everything else.
Now let a keyword $w:=A_{i_1}A_{i_2}\ldots A_{i_r}$ be given (repetitions allowed; in your case $w:=\,$ACD), and consider a random string of $N$ letters. The probability that $w$ appears in this string starting at a given position $k$ between $1$ and $N-r+1$ inclusive computes to $$p=\prod_{\ell=1}^r p_{i_\ell}\ .$$ In particular the expected number of appearances of this word at this particular place is $p$. Since this word may start at $N-r+1$ positions in all, by linearity of expectation the expected total number of appearances of $w$ is given by $$E=(N-r+1) p\ .$$