gantz-thomas-gantz / bachelorthesismathematics

Least Squares and Huber regression via CQPs

Bachelor Thesis Mathematics

The code was built during my work on my thesis for the B.Sc. Mathematics 100% at the University of Heidelberg. The objective of this thesis is to introduce a robust one-dimensional polynomial regression model using the Huber cost function via Convex Quadratic Optimisation and to compare its performance against the non-robust Least Squares approach via a concrete implementation using the qpOASES algorithm for solving the obtained Convex Quadratic Programs, which must be installed beforehand.