In: Statistics and Probability
What is bootstrap sampling? Provide advantages and disadvantages of bootstrap sampling.
Why bootstrap sampling is better than the basic simple random sampling?
Provide one real world case where bootstrap sampling is more appropriate.
The simple random sampling is where we select randomly without replacement and each of the observation has equal chance of being selected, but once selected, that particular observation cannot be selected again.
Bootstrapping is a completely different method, where the sampling takes place with replacement, therefore the initial probabilities of drawing a particular value remains the same and so the same values could be selected again and again.
Advantage: It is a simple procedure, easy to use. It is also an appropriate tool for measuring the stability of results in case of different results or data set. Bootstrapping gives asymptotically a more accurate confidence interval then the simple random sampling procedure.
Disadvantage: It generally does not provide a guarantee as it is almost impossible to know the real confidence interval under real world situation. Assumptions likes independence of samples are required here too.
One real world case where bootstrapping could be used is logistic regression analysis where the number of successes that is dependent variable, Y = 1 is very low, in this case bootstrapping could be used and the probability of getting Y = 1, could be increased by duplicating those results.