COMPSCI C8 Lecture Notes - Lecture 24: Null Hypothesis, Confidence Interval, Standard Deviation

• School: Berkeley
• Department: Computer Science
• Course: COMPSCI C8
• Professor: John Denero

Document Summary

Method: construct a (100-p)% usually 95% confidence interval for the population average. If x is not in the internal, reject the null. If (cid:454) is i(cid:374) the i(cid:374)ter(cid:448)al, (cid:272)a(cid:374)"t reje(cid:272)t the (cid:374)ull. Need not be an integer even if the data are integers. Somewhere between min and max, but not necessarily halfway in between. Same units as the data (cid:862) (cid:373)oothe(cid:374)s out (cid:448)alues(cid:863): (cid:272)olle(cid:272)t all the (cid:272)o(cid:374)tri(cid:271)utio(cid:374)s i(cid:374) o(cid:374)e (cid:271)ig plot, the(cid:374) split e(cid:448)e(cid:374)l(cid:455) Not the (cid:862)half(cid:449)a(cid:455) poi(cid:374)t(cid:863) of the data; the (cid:373)ea(cid:374) is (cid:374)ot the (cid:373)edia(cid:374) . Unless the distribution is symmetric about a appoint, then that point is both average and median. If the histogram is skewed, then the mean is pulled away from the median in the direction of the tail. Pla(cid:374) a: (cid:862)(cid:271)iggest (cid:448)alue s(cid:373)allest (cid:448)alue(cid:863: does(cid:374)"t pro(cid:448)ide i(cid:374)for(cid:373)atio(cid:374) a(cid:271)out shape of distri(cid:271)utio(cid:374) Plan b: measure variability around the mean, need to figure out a way to quantify this.