Hardness Prediction for the University Course Timetabling Problem

This paper presents an attempt to find a statistical model that predicts the hardness of the University Course Timetabling Problem by analyzing instance properties. The model may later be used for better understanding what makes a particular instance hard. It may also be used for tuning the algorithm actually solving that problem instance. The paper introduces the definition of hardness, explains the statistical approach used for modeling instance hardness, as well as presents results obtained and possible ways of exploiting them.

• Type: Conference
• Year: 2004
• Where: EVOW
• Authors: Philipp Kostuch, Krzysztof Socha
• Keywords: Artificial Intelligence, EVOW 2004, Instance Hardness, Instance Properties, Particular Instance