View Proposal

---

Proposer

Jing Wei Teoh

Title

On designing the sequential likelihood ratio test with truncation rule

Goal

Description

This project aims to explore the sequential version of the likelihood ratio test for statistical hypothesis testing. In this project, the student is expected to replicate the framework of the sequential likelihood ratio test (SLRT), develop a truncation rule for the test, and construct an algorithm to design the truncated SLRT. 𝗠𝗼𝘁𝗶𝘃𝗮𝘁𝗶𝗼𝗻: According to the Neyman-Pearson lemma, the likelihood ratio test (LRT) is the most powerful test among a class of statistical tests with fixed Type-I error rate α. The SLRT was developed to allow for a more flexible sampling strategy, i.e., instead of taking a fixed number of observations at the outset, the SLRT evaluates a decision after taking each individual observation, and stops sampling when a decision (to reject H0/not reject H0) is made. As the SLRT samples a random number of observations, its feasibility in real world applications is often concerning: practitioners may be held back if the SLRT takes 50 observations to reach a decision. The idea of truncating the SLRT is hence to address the impracticalities of the original SLRT.

Resources

1. F79MA Statistical Models A, Week 5 Reading Material 2. Schmegner, C. (2009). Truncated sequential procedures. Sequential Analysis, 28(3), 406-422. 3. Liu, Y., & Li, X. R. (2013). Performance analysis of sequential probability ratio test. Sequential Analysis, 32(4), 469-497.

Background

Url

Difficulty Level

High

Ethical Approval

InterfaceOnly

Number Of Students

1

Supervisor

Jing Wei Teoh

Keywords

Degrees

Bachelor of Science in Statistical Data Science