Ndeye Niang

Combined use of association rules mining and clustering methods to find relevant links between binary rare attributes in a large data set

A method to analyse links between binary attributes in a large sparse data set is proposed. Initially the variables are clustered to obtain homogeneous clusters of attributes. Association rules are then mined in each cluster. A graphical comparison of some rule relevancy indexes is presented. It is used to extract best rules depending on the application concerned. The proposed methodology is illustrated by an industrial application from the automotive industry with more than 80 000 vehicles each described by more than 3000 rare attributes.

Multidimensional Methods for Statistical Process Control: Some Contributions of Robust Statistics

We deal with the problem of multivariate process control. The aim of this chapter is to contribute to the development of methodological and computational tools for a simultaneous consideration of several characteristics. First a critical study of the literature about multivariate Shewhart charts will be done and two new multivariate moving average charts are proposed. Afterwards we point out a number of limitations of the Shewhart chart for the first stage of the process control. These limitations make this first stage inapplicable. In order to remedy this, a set of methods based on robust estimations of the process mean and dispersion is proposed. Simulated examples are given to illustrate the methods.

Multivariate statistical control of batch processes with variable duration

Batch processes are widely used in several industrial sectors. In those processes performance is described by variables which are monitored as the batch progresses, typically using control charts based on multiway principal components analysis (CCPs). Here we investigate the special case of batches with variable duration, which cannot be directly monitored using CCPs. We propose a new quality control strategy for monitoring such batches which are not aligned or time warped with respect to their trajectories, but are rather completed using an alternative scheme such that all information on the variability in batch profiles along the time axis is preserved. The completed data set is reduced using the Statis method and monitoring of batch performance is accomplished directly on principal plane graphs, from which non-parametric control charts are derived.

Clusterwise analysis for multiblock component methods

Multiblock component methods are applied to data sets for which several blocks of variables are measured on a same set of observations with the goal to analyze the relationships between these blocks of variables. In this article, we focus on multiblock component methods that integrate the information found in several blocks of explanatory variables in order to describe and explain one set of dependent variables. In the following, multiblock PLS and multiblock redundancy analysis are chosen, as particular cases of multiblock component methods when one set of variables is explained by a set of predictor variables that is organized into blocks. Because these multiblock techniques assume that the observations come from a homogeneous population they will provide suboptimal results when the observations actually come from different populations. A strategy to palliate this problem—presented in this article—is to use a technique such as clusterwise regression in order to identify homogeneous clusters of observations. This approach creates two new methods that provide clusters that have their own sets of regression coefficients. This combination of clustering and regression improves the overall quality of the prediction and facilitates the interpretation. In addition, the minimization of a well-defined criterion—by means of a sequential algorithm—ensures that the algorithm converges monotonously. Finally, the proposed method is distribution-free and can be used when the explanatory variables outnumber the observations within clusters. The proposed clusterwise multiblock methods are illustrated with of a simulation study and a (simulated) example from marketing.

Controle multivariado de processos em batelada com duração variável

Batch processes are widely used in several industrial sectors, such as food and pharmaceutical manufacturing. In a typical batch, raw materials are loaded in the processing unit and submitted to a series of transformations, yielding the fi nal product. Process performance is described by variables which are monitored as the batch progresses. Data arising from such processes are likely to display a strong correlation-autocorrelation structure, and are usually monitored using control charts based on multiway principal components analysis (MPCA charts). In this paper we investigate the special (and rather frequent) case of batches with varying duration, which cannot be directly monitored using MPCA charts. We propose a new quality control strategy for monitoring such batches. In our proposition, batches are not aligned or time warped with respect to their trajectories, but are rather completed using a straightforward scheme. Thus all information on the variability in batch profi les along the time axis is preserved. The data set completed is reduced using the Statis method and monitoring of batch performance is accomplished directly on principal plane graphs, from which non-parametric control charts are derived. A simulated example illustrates the proposed method.