Bachelor's Degree Project Presentations in Mathematics: Mira Åkerman, Lukas Billera and Kevin Haagensen Strömberg

Mira Åkerman

Time: 10:15-11:00

Place: 80115

Title: Comparing of Imputation Methods for Missing Data in Logistic Regression

Abstract: This presentation examines how different imputation methods influence coefficient estimates in a logistic regression model. A complete dataset was first used to fit a base model using logistic regression. Missing data was then introduced through simulations with different proportions of missing values. Both single and multiple imputation methods were applied to the incomplete datasets. The coefficient estimates obtained after imputation were compared with those from the base model using the L1 norm. This comparison was used to evaluate how well the different imputation methods preserved the base model estimates.

Lukas Billera

Time: 14:15-15:00

Place: 80115

Title: Reward-tilted sampling for diffusion generative models and Branching Flows

Kevin Haagensen Strömberg

Time: 15:15-16:00

Place: 80115

Title: How to Solve Your Hat Problem

Abstract: We analyse various hat problems from the perspective of problem solving. The problems considered are the standard hat problem as well as variations involving multiple guesses, hats in boxes and unknown colours. The goal is to understand how one comes up with their solutions and which ideas are important for the solutions. The main result is motivating the technique of using the axiom of choice. The technique appears as a representation of common knowledge among agents which in turn arises from a need for coordination.