Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
16 Cards in this Set
- Front
- Back
Statistics inference
|
A process that we can get information and draw conclusion about the population from the sample
|
|
Estimation
|
Estimate the parameters of the population using the statistics.
x_bar -> mu; s -> sigma |
|
Point estimator (PE)
|
Draw inference for an unknown parameter about a single value or a point
|
|
Drawbacks about PE
|
(1) For continuous probability, P(x=a)=0 implies we hava zero prob. to have x_bar = mu, for example.
(2) Most of the time, we want how close the statistics to the parameters. PE cannot provides this info. (3) As sample size (ss) increase PE won't be affect. As a good estimator, if ss increases, it will get more accurate. |
|
Interval estimator
|
Draw inference for the unknown parameter about an interval.
i.e. within that interval, the parameter will be included with x% of confidence. |
|
What is are good qualities for the estimator?
|
(1) unbiased
(2) consistency (3) relatively efficiency |
|
unbiased
|
expected value of an estimator is equal to the parameter
|
|
consistency
|
as sample size increase, the difference btw an estimator and the parameter will be decrease
|
|
relative efficiency
|
If there are 2 unbiased estimator, the one with smaller variance is relative efficiency.
|
|
CI of mu, with CI = 1-alpha
|
CI=[x_bar-za_2x(sigma/r_n),
x_bar+za_2x(sigma/r_n)] interpret: the probability that mu as a fixed value included in the CI is 1-alpha |
|
lower/upper CI limit
|
LCI = lower limit of CI
UCI = upper limit of CI |
|
Useful z_a_over_2 values:
90%=? 95%=? ... |
1-a z_a_over_2
90% 1.645 95% 1.96 98% 2.33 99% 2.575 |
|
CI interpretation
|
About CI% of time for the repeatly sampling till infinity time that the true parameter is included within CI
Since in reality, we pick only one sample usually, our statistics and we only know that if we repeatly draw the samples we will have CI% of the samples contain the parameter |
|
CI is affected by ___, ___, ___
|
(1) 1-alpha i.e. confidence level; inc->narrow
(2) n; inc->narrow (3) sigma; inc-> wider the wider CI is, the less precisous info it produces |
|
sigma of sample median
|
= 1.2533sigma/root_n
=> bigger than sample mean, less relative efficiency |
|
Sample size
|
If width of CI and CI% is specific, and sigma is given then
w = z_a_2*[sigma/root_n] => n=((z_a_2*sigma)/w)^2 Give sample size, we then can get a sample and find sample mean; Then, find CI. |