-2 longer apply Hoeffding's inequality, and can only use Chebyshev's inequality. This is a problem if we require a very high confidence level: in this part, you'll show why. For this part only, assume that o' = 1, a = -1,b = 1, and e = 0.1. Make a plot that shows, for particular values of 8, the number of samples n required based on your answers from parts (a) and (b). Your plot should show a range of ten o values between 1/2 and 1/1000, using np . geomspace (1/2, 1/1000, 10), and should be shown on a log-log scale. What do you observe