Segment 14

From Computational Statistics (CSE383M and CS395T)
Jump to navigation Jump to search

Calculation Problems

1. P-value to rule out the hypothesis that a coin is fair when 8 out 10 flips were heads. Is this statistically significant?

p value = <math>\frac{(1 + 10 +45 + 45 + 10 + 1)}{2^{10}} = 0.109375</math> which is not statistically significant.

2. As a Bayesian if the prior that the coin is fair is 0.75, what is the posterior prior that the coin is fair?

If the event of the coin being fair be <math>K_0</math> and that the coin is unfair be <math>K_1</math>. Assuming that being a fair and an unfair are mutually exclusive and exhaustive, the probability of <math>K_1 = 0.25</math>

<math> \begin{align} P(K_0|Data) &= \frac{P(Data|K_0)\cdot P(K_0)}{P(Data|K_0)\cdot P(K_0) + P(Data|K_1)\cdot P(K_1)}\\ &= \frac{\left(\frac12 \right)^{10} \cdot 0.75}{\left(\frac12 \right)^{10} \cdot 0.75 + \int_0^1 p^2(1-p)^8 dp}\\ &= 0.591 \end{align} </math>


Food for Thought Problems

Class Activity