casino 168

The thesis consists of four papers dealing with the design or application of supervised classification with mixed data. Using generalized linear models, the first two papers suggest methodologies to reduce the number of coefficients associated with categorical predictors and their interactions. In the third paper, predictive modeling is applied to the problem of school dropout in Colombia, and strategies are suggested for targeting at-risk students. Using feature shrinkage, the fourth paper proposes a method to enhance the trade-off between accuracy and unfairness in supervised learning.

Ìý

Primary Supervisor:
Professor Dolores Romero Morales

Department of Economics

Copenhagen Business School

Ìý

Secondary Supervisors:

Professor Emilio Carrizosa

Instituto de Matemáticas

Universidad de Sevilla

Ìý

Associate Professor Fane Groes

Department of Economics

Copenhagen Business School

Ìý

Assessment Committee:

Associate Professor Moira Daly (Chair)

Department of Economics

Copenhagen Business School

Ìý

Associate Professor Jochen De Weerdt

Faculty of Economics and Business

Katholieke Universiteit Leuven

Ìý

Full Professor Sebastian Maldonado

Universidad de Chile

Ìý

Thesis:
The thesis will be available from

Ìý

Reception:
The casino 168 PhD School will host a reception, which will take place immediately after the defence in ECONS Kitchen at ±Ê´Ç°ù³¦±ð±ôæ²Ô²õ³ó²¹±¹±ð²Ô (3th. Floor).

Ìý