COMPSCI C8 Lecture Notes - Lecture 32: Scatter Plot, Null Hypothesis, Confidence Interval
Document Summary
Distance drawn at random from normal distribution with mean 0. Another distance drawn independently from the same normal distribution. Find the slope of the regression line through the bootstrapped plot. Draw the empirical histogram of all the generated slopes. That"s a(cid:374) appro(cid:454)i(cid:373)ate 95% co(cid:374)fide(cid:374)ce i(cid:374)terval for the slope of the true line. Null hypothesis: the slope of the true line is 0. Method: construct a bootstrap confidence interval for the true slope. If the i(cid:374)terval does(cid:374)"t co(cid:374)tai(cid:374) 0, reject the (cid:374)ull hypothesis. If the i(cid:374)terval does co(cid:374)tai(cid:374) 0, there is(cid:374)"t e(cid:374)ough evide(cid:374)ce to reject the (cid:374)ull hypothesis.