- Independent: probability of A does not effect probabiliy of B
$P(X\leq x,Y\leq y)=P(X\leq x)P(Y\leq y)$
- identically distributed: Events follow the same distribution. 两个事件遵循同样的分布
PMF&PDF:
CDF:
PMF: for discrete
PDF: for continuous
CDF: cumulated posibility
CDF = PMF/PDF's integral
PMF/PDF = CDF's derivation
离散型 For discrete:
$E(X) = \sum^{\infin}{j=1} x_j P(X=x_j)$
$E(X) = \sum^{\infin}{k=1} x_kp_k$
连续型 For continuous:
The variance of an r.v. X is
For discrete:
For continuous:
Covariance:
- sampling with replacemen:
$Var(\overline{X}) = \frac{1}{n^2} \sum_{i=1}^nVar(X_i) = \frac{\sigma^2}{n}$ - sampling without replacement:$Var(\overline{X}) = \frac{\sigma^2}{n}(1-\frac{n-1}{N-1}) \approx \frac{\sigma^2}{n}$
-
$H_0$ Null hypothesis -
$H_1$ Alternate hypothesis $H_0 = H_1^c, P(H_0) + P(H_1) = 1$
Terminology | Meaning |
---|---|
Type I error | Rejecting |
Significance leve |
Probability of a type I error |
Type II error | Accepting |
The power of test |
Probability that |
- Find 2 events A and B
- Find all elements in LOTP
Use:
Especially:
PMF: $P(X = k) = \begin{pmatrix} n \ k \end{pmatrix} p^k(1-p)^{n-k}$
Likelihood ration for
threshold c
For X = c
p-value =
PDF:
Valid PDF => CDF from a to b is 1
For x:
For all x with
Confidence intervals: Lower endpoint:
- Population mean
- Population total N
- Population variance
- The sample size (usually, depends on how to choose a sample)
- significant level
- Sample size n
- The variance of the sample mean
- The estimated variance of the sample mean
- Sample mean
$\overline{X}$ - The largest value in the sample
- The interval end points
- Find
$f(x_i|\theta)$
e.g.: n coins, S are Head, n-S are tails $f(x_i|\theta) = (ppp...p)((1-p)(1-p)...) = p^S(1-p)^{n-S}$ - log-likehood:
e.g.:$f(x_0|\theta) = p^S(1-p)^{n-S}$ $l_n(p) = log(p^S(1-p)^{n-S}) = log(p^S) + log((1-p)^{n-S}) = S\log{p} + (n-S)\log {(1-p)}$ - Find maximum value (derivative = 0)
$l_n'(p) = S/p - \frac{n-S}{1-p} = 0$ - Find p:
$p = \frac{S}{n}$ - The MLE is
$\hat{p} = \frac{S}{n}$