Can anyone help me with this question
If the real answer is "no", then give 50% yes and 50% no If the real answer is "yes“, then give 60% yes and 40% no
Suppose that the noisy answers consist of 55% yes and 45% no Estimate the percentage of "yes" in the original answer
This was my attempt
$p(\text{yes | the real answer is "no"}) = 0.5$
$p(\text{no | the real answer is "no"}) = 0.5$
$p(\text{yes| the real answer is "yes"}) = 0.6$
$p(\text{no | the real answer is "yes"}) = 0.4$
$p(\text{yes}) = 0.55$
$p(\text{no}) = 0.45$
p(the real answer is "yes | yes) = p(yes| the real answer is "yes") / [p(yes| the real answer is "no") + p(yes| the real answer is "yes")] $= 0.6 / [0.5 + 0.6] = 0.54$
p(the real answer is "yes ^ yes) = p(the real answer is "yes | yes) x p(yes) $= 0.54 \times 0.55 = 0.3$
I feel what I am doing is wrong.
is $p(A|B) = p(B|A)/[p(B|A) + p(B|\text{not }A)]$ ?
Let $X$ be the event the real answer is "yes" and $Y$ be the event the noisy answer is "yes". The question tells you:
Since $$\mathbb P(Y) = \mathbb P(Y \mid X)\mathbb P(X)+ \mathbb P(Y \mid X^c)\mathbb P(X^c)$$ i.e. $$\mathbb P(Y) = \mathbb P(Y \mid X)\mathbb P(X)+ \mathbb P(Y \mid X^c)(1-\mathbb P(X))$$ you can say
$$\mathbb P(X)=\frac{\mathbb P(Y)- \mathbb P(Y \mid X^c)}{\mathbb P(Y \mid X)-\mathbb P(Y \mid X^c)}$$
which in this case is $\dfrac{0.55-0.5}{0.6-0.5}=0.5$