M.P. Derde

UNEQ: a disjoint modelling technique for pattern recognition based on normal distribution

Abstract UNEQ, a method for supervised pattern recognition based on the assumption of multivariate normally-distributed groups, is presented. The method belongs to the group of so-called class-modelling techniques, i.e., classification functions are developed for each of the training classes separately, on the basis of the similarities between the objects within a group. New classes can therefore be entered easily into a classification problem. The method allows also easy detection of outliers. For each individual sample, the degree of connection with all the training classes can be defined. If for a given sample, this degree of class membership is low for all the classes, the object is considered as an outlier. The mathematical background of UNEQ is described. The validation of the derived classification functions in terms of sensitivity, specificity and efficiency is discussed. The method is illustrated and compared to SIMCA (another class-modelling technique) by means of a data set that concerns the classification of olive oils according to their area of origin, based on fatty acid patterns. It is concluded that the UNEQ method can be very useful for classification purposes but requires the populations to be homogeneous as is the case for other techniques. For the olive-oil data set, the performance of UNEQ is similar to or even better than SIMCA.

Use of a microcomputer for the definition of multivariate confidence regions in medical diagnosis based on clinical laboratory profiles

The use of multivariate confidence regions is proposed for the classification of clinical laboratory profiles into diagnostic classes. For this purpose, a multivariate confidence region is developed for each diagnostic class. Three methods (UNEQ, EQ, and SIMCA) are evaluated and compared with classical linear discriminant analysis. As an example, a small data set concerning the differentiation of the thyroid functional states on the basis of five laboratory tests is used. It is shown that related procedures can produce results of very different quality and that the multivariate region approach is attractive for the clinicians daily practice since the methods are easily implemented on a microcomputer.

Supervised pattern recognition: the ideal method?

Abstract The different criteria that should be considered in selecting a supervised pattern recognition technique for a particular application are discussed. An overview is given of the most important and most frequently-used supervised techniques and the extent to which they meet the criteria. The possibilities of two rule-building expert systems are also discussed.

Effect of scaling on class modeling with the simca method

Abstract The first step in multivariate analysis is almost always the scaling of the variables. The pattern recognition technique SIMCA provides the possibility of scaling the variables over all the objects of the training set (classical scaling), or only over the objects belonging to the same group (separate scaling). The former method of scaling is the more used. The effect of separate scaling on the classification of objects with SIMCA is investigated for a data set consisting of the percentage distribution of fatty acids in olive oils originating from two neighbouring regions in Italy. It is shown that separate scaling has a beneficial effect on the classification.

Comparison of the performance of the class modelling techniques UNEQ, SIMCA, and PRIMA

Abstract Derde, M.P. and Massart, D.L., 1988. Comparison of the performance of the class modelling techniques UNEQ, SIMCA, and PRIMA. Chemometrics and Intelligent Laboratory Systems, 4: 65-93 By means of a Monte Carlo study a systematic comparison of the supervised pattern recognition techniques of the class modelling type, UNEQ, SIMCA and PRIMA is made. In particular, the success rate of the classification decisions and the influence of the sample size on it were investigated. It was concluded that better class models are obtained when a technique is used that takes the shape of the population distribution into account. If the actual distribution cannot be determined, then use should be made of techniques that make no or only weak assumptions about the shape of the distribution. However, even then it remains worthwhile to investigate whether the variables are correlated and to take this information into account. When using SIMCA and PRIMA, attention should also be paid to the way the class models are defined: an approach that makes use of certain sample parameters such as the range of the variables or the maximum distance between a training object and the class model might lead to overly broad models, especially for large training sets.

Extraction of information from large data sets by pattern recognition

The use of computers has made data collection much easier and analytical chemists increasingly wonder how to make use of all the data obtained. Pattern recognition permits to extract information present in large data sets in an automatic way.Many scientists acknowledge this fact but are rebutted by the task of learning to use pattern recognition methods. Indeed, there are many methods available and for the newcomer it is extremely difficult to make a selection. For this reason, the lecture will start by explaining the models used in pattern recognition. This will be followed by a critical discussion of advantages and disadvantages of the methods and a selection of preferred methods.

Potential methods in pattern recognition : Part 3. Feature selection with ALLOC

Abstract The feature selection procedure of ALLOC is compared with the SELECT procedure in the ARTHUR software package and with a procedure based on statistical tests in the SPSS software package. Since ALLOC classification is very sensitive to redundant variables, feature selection is necessary. This is not a disadvantage because detection of redundant variables is always desirable. The ALLOC selection procedure performs very well in the two applications considered here, i.e., differentiation of milk samples and characterization of thyroid function.

Evaluation of the required sample size in some supervised pattern recognition techniques

Abstract One of the aspects of supervised pattern recognition applications which is often ignored concerns the minimum number of samples necessary to define sufficiently reliable classification rules. For discriminating techniques, criteria are available that can be used to evaluate this minimum sample size. For class-modelling techniques, however, no attention has previously been paid to this aspect. Here, the guidelines that can be applied in the use of a discriminating supervised technique are discussed, and criteria are proposed that can be applied when class-modelling supervised techniques, particularly UNEQ, are applied.

UNEQ: A class modelling supervised pattern recognition technique

The purpose of this study is to compare the performance of UNEQ, a supervised pattern recognition technique of the classmodelling type, with other classification techniques and to demonstrate its use for a practical application, namely the classification of coals.

Feasibility study concerning the use of expert systems for the development of procedures in pharmaceutical analysis

The feasibility of using expert systems for the development of analytical procedures is investigated. A system for the computer generation of procedures to determine active drug substances in commercial formulations is proposed. It is shown that in nearly 85% of the cases investigated the present system immediately yields a correct procedure or conclusion. It is concluded that selecting methods and developing procedures with the use of expert systems is difficult but feasible.