Abstract
This course teaches modern statistical inference methods known as Bootstrap, with various real-world examples of statistical problems, such as testing whether a new medical treatment is effective in curing disease. One main aim of statistical inference is to quantify the error of a statistical estimator for a population quantity, such as the difference between the health outcomes of treated and controlled patients. Error quantification is needed to obtain a valid conclusion about the treatment effect. Clasical statistics generally rely on an mathematical approximation of the distribution of a statistical estimate. Modern statistics based on Bootstrap do not make such an approximation; instead, it estimates the distribution of the statistical estimate by simulating the population distribution by resampling from the given data. This Bootstrap approach has widely made statistical inferences more accurate, more widely applicable, and easier to use and understand. The course not only teaches various aspects of the Boostrap approach, but also serves as a practical introduction to statistics applied to real-world problems through examples.
Teaching and Learning Methods: Lectures, simulation experiments, and homework.
Course Policies: The class attendance may be taken into account in the final grade.
Book: BRADLEY E., ROBERT J. T. An Introduction to the Bootstrap. CRC Press, 1994
Prerequisites
Understanding basic statistical concepts, such as the bias and variance of a statistical estimator.
The agenda of the course is as follows:
- The basic idea of bootstrap
- The bootstrap method for estimating standard errors
- Applications to various statistical problems
- Bootstrap for regression
- Estimation of bias
- Hypothesis testing with the bootstrap
Learning Outcomes: Use the Bootstrap method to:
- Perform statistical inference, such as calculating confidence intervals and hypothesis testing, in variousproblems.
Nb hours: 21.00
Evaluation:
- Homework (25% of the final grade)
- Exam (75% of the final grade)