Research team
Expertise
Statistical consulting, including: -data analysis by means of statistical software -development of algorhythms and programs for regression, prediction, quality control, clustering, image processing, etc.
Development of time-efficient algorithms for depth functions based on techniques of computational geometry, and new applications to economic data.
Abstract
The first goal is to further investigate the notion of location depth, and to construct time-efficient algorithms to compute the depth and the corresponding contours and location estimator for a given data set in 2, 3, or more dimensions. It will also be attempted to extend this work to the equally important problem of estimating the scatter structure of the data. The second goal is to analyze and model economical systems, like inflation and trading on financial markets, by means of depth functions and other robust techniques.Researcher(s)
- Promoter: Rousseeuw Peter
- Fellow: Struyf Anja
Research team(s)
Project type(s)
- Research Project
Numerical and Monte Carlo algorithms for the pricing of exotic options.
Abstract
Researcher(s)
- Promoter: Rousseeuw Peter
- Fellow: Symens Stijn
Research team(s)
Project type(s)
- Research Project
Robust Multivariate Methods and Financial Statistics.
Abstract
Since classical multivariate statistical methods cannot resist outliers, we construct robust multivariate methods and investigate their properties such as local and global robustness, consistency and efficiency. We will also study financial models. Different parameter estimation techniques will be investigated and compared. We will investigate the consequences for the models if some of the model assumptions are not satisfied.Researcher(s)
- Promoter: Rousseeuw Peter
- Fellow: Van Aelst Stefan
Research team(s)
Project type(s)
- Research Project
Statistical and numerical techniques for the modelling and optimization of computer- and communication networks
Abstract
This project will analyze the performance of advanced technological systems such as communication networks (including the Internet), computer systems, and distributed multiprocessor systems, with the aim of optimizing their design and dimensions. This analysis will use probabilistic models, the parameters of which will be obtained by statistical estimates based on measurements of actual traffic. The computation of the performance functions and the design of optimal networks both lead to complex computational problems, which will be approached by si mulation and by novel numerical analysis techniques such as multivariate rational approximations. The many interactions between all these aspects require an intensive collaboration between the three research groups.Researcher(s)
- Promoter: Cuyt Annie
- Co-promoter: Blondia Chris
- Co-promoter: Rousseeuw Peter
Research team(s)
Project type(s)
- Research Project
Computational Methods for Performance Evaluation and Simulation of Complex Technical Systems
Abstract
Starting from observed data (like traffic on the Internet), robust statistical methods (i.e., techniques that give reliable results even when deviations occur in the input data) will be applied to construct a model for the observed system. From the specific architecture and structure of the system one can often derive interesting properties of the performance measure in advance, such as its monotonicity relative to a given system parameter, or its asymptotic behavior. These properties are helpful when constructing the performance measure, but by themselves they are not sufficient. Robust, efficient and accurate approximations of the exact solution are indispensable. A possible approach is based on power series, but since many performance functions have singularities a better approach is to use Pade approximations.Researcher(s)
- Promoter: Blondia Chris
- Co-promoter: Cuyt Annie
- Co-promoter: Rousseeuw Peter
Research team(s)
Project type(s)
- Research Project
Investigation of nominal rigidities in Belgian consumer prices.
Abstract
Inflation is often measured by the index of consumer prices. This is a weighted average of prices of household products. Each month the inflation is estimated from the price increments, at a certain aggregation level (i.e., 60 product categories). The empirical distribution of these increments is often skewed, and may contain outliers. To this end, robust statistical techniques will be developed, also taking into account the distinction between `sticky prices' (which only change from time to time) and flexible prices (like gasoline or heating oil).Researcher(s)
- Promoter: Rousseeuw Peter
Research team(s)
Project type(s)
- Research Project
Robust estimators of covarance matrices.
Abstract
We intend to construct faster algoritms for robust estimators in the case of multivariate analysis. Examples are the minimum volume ellipsoid and the class of S-estimators. We will also investigate their distribution for finite samples, which we need for the construction of tests and confidence regions around the estimated parameters.Researcher(s)
- Promoter: Rousseeuw Peter
- Fellow: Van Aelst Stefan
Research team(s)
Project type(s)
- Research Project
Development of new and efficient algorithms for data analysis and data mining.
Abstract
We want to construct techniques with high breakdown value, high efficiency and feasible computation time. For this purpose the concept of location and regression depth will be generalized to the depth of scatter matrices. Based on these scatter estimators we want to develop clustering methods. We will extend the results to factor analysis and discriminant analysis.Researcher(s)
- Promoter: Rousseeuw Peter
- Fellow: Hubert Mia
Research team(s)
Project type(s)
- Research Project
Algorithms for fuzzy classification.
Abstract
Automatic classification is the division of data into homogeneous subsets. This discipline belongs to computer science, pattern recognition, and statistics, with many applications. The goal of this project is the development of classification techniques which are robust towards the usual model assumptions. The research primarily concerns fuzzy set techniques, with a particular interest towards new methods with high contrast.Researcher(s)
- Promoter: Rousseeuw Peter
- Fellow: Struyf Anja
Research team(s)
Project type(s)
- Research Project
Computational methods for performance evaluation and simulation of complex technical systems.
Abstract
The analysis and performance evaluation of advanced technical systems, such as computer systems, telecommunication systems and distributed multiprocessor systems often involve solving a complex computational problem. This is due to the fact that the complexity increases with the size of the system under study or with the dimensions of the system model. This implies that it is preferable to use rational functions rather than polynomials.Researcher(s)
- Promoter: Blondia Chris
- Co-promoter: Cuyt Annie
- Co-promoter: Rousseeuw Peter
Research team(s)
Project type(s)
- Research Project
Robust estimators of covarance matrices.
Abstract
We intend to construct faster algoritms for robust estimators in the case of multivariate analysis. Examples are the minimum volume ellipsoid and the class of S-estimators. We will also investigate their distribution for finite samples, which we need for the construction of tests and confidence regions around the estimated parameters.Researcher(s)
- Promoter: Rousseeuw Peter
- Fellow: Van Aelst Stefan
Research team(s)
Project type(s)
- Research Project
Algorithms for fuzzy classification.
Abstract
Automatic classification is the division of data into homogeneous subsets. This discipline belongs to computer science, pattern recognition, and statistics, with many applications. The goal of this project is the development of classification techniques which are robust towards the usual model assumptions. The research primarily concerns fuzzy set techniques, with a particular interest towards new methods with high contrast.Researcher(s)
- Promoter: Rousseeuw Peter
- Fellow: Struyf Anja
Research team(s)
Project type(s)
- Research Project
Computational statistics : new methods and time-efficient algorithms.
Abstract
For scientific computations based on multivariate observations we develop new methods which yield reliable results even when outliers occur. Constructing fast algorithms for these methods is not trivial because of the inherent combinatorics and other computer-intensive aspects.Researcher(s)
- Promoter: Rousseeuw Peter
Research team(s)
Project type(s)
- Research Project
Modern Computational Methods in Applied Mathematics.
Abstract
Nowadays many new methods in applied mathematica are very computational intensive, e.g. wavelets in numerical analysis and statistics, genetic algorithms, cluster analysis, detection of multivariale outliers, and the construction of depth functions. This project aims to investigate these methode and to develop faster algorithms.Researcher(s)
- Promoter: Wuytack Luc
- Co-promoter: Rousseeuw Peter
Research team(s)
Project type(s)
- Research Project
Algorithms for fuzzy classification.
Abstract
Automatic classification is the division of data into homogeneous subsets. This discipline belongs to computer science, pattern recognition, and statistics, with many applications. The goal of this project is the development of classification techniques which are robust towards the usual model assumptions. The research primarily concerns fuzzy set techniques, with a particular interest towards new methods with high contrast.Researcher(s)
- Promoter: Rousseeuw Peter
- Fellow: Struyf Anja
Research team(s)
Project type(s)
- Research Project
Chemometrics.
Abstract
In chemometrics, statistical and other quantitative techniques are applied to chemical analysis. Much use is made of pattern recognition, optimalisation, multivariate modelling, cluster analysis, and robust methods. Our contribution to this project is mainly in the latter two topics.Researcher(s)
- Promoter: Rousseeuw Peter
Research team(s)
Project type(s)
- Research Project
Robust estimators in multivariate statistical models.
Abstract
In robust statistics one uses methods which also work under deviations from the proposed distribution (where a lot of classical methods fail). We are looking for estimators in the regression model and the multivariate location/scatter model which combine robustness (high breakdown point, low biascurve), statistical efficiency and which are rather fast to compute. These estimators will be implemented and tested on real and simulated examples.Researcher(s)
- Promoter: Rousseeuw Peter
- Fellow: Croux Christophe
Research team(s)
Project type(s)
- Research Project
An ABC-study for the motor car assessment service.
Abstract
By order of royal association of motor car assessors in Belgium a product costing analysis is undertaken in order to determine the cost drivers which influence the cost of a motor car assessment. With the use of overhead value analysis, the different cost and activity drivers are identified.Researcher(s)
- Promoter: Jorissen Ann
- Co-promoter: Reyns Carl
- Co-promoter: Rousseeuw Peter
- Co-promoter: Van Wouwe Martine
Research team(s)
Project type(s)
- Research Project
Vision.
Abstract
Researcher(s)
- Promoter: Van Dyck Dirk
- Co-promoter: Dommisse Roger
- Co-promoter: Jacob Willem
- Co-promoter: Lowen Bob
- Co-promoter: Raman Erik
- Co-promoter: Rousseeuw Peter
- Co-promoter: Van Espen Piet
- Co-promoter: Van Landuyt Joseph
- Co-promoter: Verschoren Alain
Research team(s)
Project type(s)
- Research Project
Efficient regression estimators with high breakdown point
Abstract
Efficient regression estimators with high breakdown point and high statistical efficiency.Researcher(s)
- Promoter: Rousseeuw Peter
- Fellow: Croux Christophe
Research team(s)
Project type(s)
- Research Project
Abstract
Researcher(s)
- Promoter: Cuyt Annie
- Co-promoter: Rousseeuw Peter
Research team(s)
Project type(s)
- Research Project