Bootstrap Confidence Interval

Problem

Compute a bootstrap confidence interval for the mean. Call random.seed(seed) and draw num_samples bootstrap resamples of size n with replacement using random.randrange(n), recording each resample mean.

Sort the means, then take the values at indices int((1 - confidence) / 2 * num_samples) and int((1 + confidence) / 2 * num_samples), clamping both to num_samples - 1. Return [lower, upper]. Empty data returns [0.0, 0.0].

Hint

Build the sorted list of resample means, then index it at the two percentile positions.

Test cases

case 1

in: [[1,2,3,4,5,6,7,8],200,0.95,42]

expected: [3,6.25]

case 2

in: [[10,12,11,13,12],100,0.9,7]

expected: [10.8,12.2]

case 3

in: [[10,20,30],50,0.8,1]

expected: [16.666666666666668,26.666666666666668]

case 4

in: [[5,5,5,5],20,0.5,0]

expected: [5,5]

Starter codePython

import random
def bootstrap_mean_ci(data, num_samples, confidence, seed):
    # Returns [lower, upper]
    # Your code here
    pass

4 test cases · run your code against them in the editor