- As part of the Completion Exercise for the Master of Science in Statistical Science, you may write and present your Master's Thesis. This oral examination is administered by your Master's Committee. Students choosing to defend a thesis should begin work on their research as early as possible, preferably in their second semester or summer of their first year in the program. Please give yourself enough time to write your thesis. Your thesis advisor (chair of your committee) should approve your thesis title. The work has to be approved by all members of your committee.
ALL STUDENTS CHOOSING TO DO A THESIS SHOULD SUBMIT A THESIS PROPOSAL (NOT MORE THAN 2 PAGES) TO THE MS DIRECTOR BY OCTOBER 15 (THIRD SEMESTER). THE THESIS PROPOSAL SHOULD INCLUDE A TITLE (TENTATIVE THAT CAN BE REFINED LATER), A LIST OF THREE COMMITTEE MEMBERS (TWO SHOULD BE FROM STATISTICS INCLUDING THE CHAIR) AND A DESCRIPTION OF YOUR WORK.
NOTE: THE MASTERS THESIS COMMITTEE SHOULD BE FORMED AND APPROVED BY GRADUATE SCHOOL AT LEAST 30 DAYS PRIOR TO YOUR DEFENSE.
For details, see the document below.
The Thesis consists of a detailed written report on a project approved by the M.S. Director, covering aspects of your contribution to the project area:
- summary of contributions and results
- discussion of open questions
- bibliographic material
The Master's Thesis and its submission must conform to the Duke University Graduate School M.S. thesis requirements.
All students choosing to do Master's Thesis should follow the steps outlined in the MSS Thesis Defense Process document.
- Hierarchical Signal Propagation for Household Level Sales in Bayesian Dynamic Models
- Logistic Tree Gaussian Processes (LoTgGaP) for Microbiome Dynamics and Treatment Effects
- Bayesian Inference on Ratios Subject to Differentially Private Noise
- Multiple Imputation Inferences for Count Data
- An Euler Characteristic Curve Based Representation of 3D Shapes in Statistical Analysis
- An Investigation Into the Bias & Variance of Almost Matching Exactly Methods
- Comparison of Bayesian Inference Methods for Probit Network Models
- Differentially Private Counts with Additive Constraints
- Multi-Scale Graph Principal Component Analysis for Connectomics
- MCMC Sampling Geospatial Partitions for Linear Models
- Bayesian Dynamic Network Modeling with Censored Flow Data
- An Application of Graph Diffusion for Gesture Classification
- Easy and Efficient Bayesian Infinite Factor Analysis
- Analyzing Amazon CD Reviews with Bayesian Monitoring and Machine Learning Methods
- Missing Data Imputation for Voter Turnout Using Auxiliary Margins
- Generalized and Scalable Optimal Sparse Decision Trees
- Construction of Objective Bayesian Prior from Bertrand’s Paradox and the Principle of Indifference
- Rethinking Non-Linear Instrumental Variables
- Clustering-Enhanced Stochastic Gradient MCMC for Hidden Markov Models
- Optimal Sparse Decision Trees
- Bayesian Density Regression with a Jump Discontinuity at a Given Threshold
- Forecasting the Term Structure of Interest Rates: A Bayesian Dynamic Graphical Modeling Approach
- Testing Between Different Types of Poisson Mixtures with Applications to Neuroscience
- Multiple Imputation of Missing Covariates in Randomized Controlled Trials
- A Bayesian Strategy to the 20 Question Game with Applications to Recommender Systems
- Applied Factor Dynamic Analysis for Macroeconomic Forecasting
- A Theory of Statistical Inference for Ensuring the Robustness of Scientific Results
- Bayesian Inference Via Partitioning Under Differential Privacy
- A Bayesian Forward Simulation Approach to Establishing a Realistic Prior Model for Complex Geometrical Objects
- Two Applications of Summary Statistics: Integrating Information Across Genes and Confidence Intervals with Missing Data