PubMed Health⌕ Search

Biomedical subjects

M Forina

Publications and source records attributed to M Forina.

7 recordsLinked to original sources
PubMed2004

Selection of useful predictors in multivariate calibration.

Ten techniques used for selection of useful predictors in multivariate calibration and in other cases of multivariate regression are described and discussed in terms of their performance (ability to detect useless predictors, predictive power, number of retained predictors) with real and artificial data. The techniques studied include classical stepwise ordinary least-squares (SOLS), techniques based on the genetic algorithms, and a family of methods based on partial least-squares (PLS) regression and on the optimization of the predictive ability. A short introduction presents the evaluation strategies, a description of the quantities used to evaluate the regression model, and the criteria used to define the complexity of PLS models. The selection techniques can be divided into conservative techniques that try to retain all the informative, useful predictors, and parsimonious techniques, whose objective is to select a minimum but sufficient number of useful predictors. Some combined techniques, in which a conservative technique is used to perform a preliminary selection before the use of parsimonious techniques, are also presented. Among the conservative techniques, the Westad-Martens uncertainty test (MUT) used in Unscrambler, and uninformative variables elimination (UVE), developed by Massart et al., seem the most efficient techniques. The old SOLS can be improved to become the most efficient parsimonious technique, by means of the use of plots of the F-statistics value of the entered predictors and comparison with parallel results obtained with a data matrix with random data. This procedure indicates correctly how many predictors can be accepted and substantially reduces the possibility of overfitting. A possible alternative to SOLS is iterative predictors weighting (IPW) that automatically selects a minimum set of informative predictors. The use of an external evaluation set, with objects never used in the elimination of predictors, or of "complete validation" is suggested to avoid overestimate of the prediction ability.

Journal Article↗
PubMed1998

Multivariate calibration: applications to pharmaceutical analysis.

The principles of multivariate calibration (MC) are presented, with reference to the main objectives of this chellometrics technique: the reduction of the variance in the prediction of a response variable (generally, a chemical quantity) and the possibility of the determination of the response in complex matrices with no or limited sample preparation, as in the case of the determination of a drug in a medicament. In both cases MC uses the whole information in a spectrum (a series of predictors). The possibility of the improvement of the MC performances, eliminating some useless, noisy, predictors is shown. Variable selection has been performed using two original techniques: a stepwise elimination procedure, based on the normalised coefficients of the regression equation relating the response to the predictors and a technique based on iterative repetitions of the regression technique (partial least squares regression, PLS), each time by weighting the predictors by their normalised regression coefficient computed in the previous cycle. These strategies are illustrated by means of different data sets, a synthetic example and a real example where MC, applied to near infrared spectroscopy, is used in the analysis of a drug. In this case also the application of an original MC technique is shown, where a joint regression model is obtained for two different instruments.

Calibration↗
PubMedDate not supplied

Cluster analysis: significance, empty space, clustering tendency, non-uniformity. I--Statistical tests on the significance of clusters.

The agglomerative clustering methods and the tests usually applied to evaluate the significance of clusters are critically evaluated. Many clustering techniques can provide erroneous information about the existence of clusters. The single linkage technique is suggested to identify natural, well separated, clusters. The existing statistical tests on the significance of clusters are not satisfactory. A new statistical test, based on the distribution of the distances between the objects and their first nearest neighbor, is presented. The performances of the test are compared with those of the Sneath test and of the variance-ratio test on some artificial and real data sets.

Chemistry Techniques, Analytical↗
PubMedDate not supplied

Cluster analysis: significance, empty space, clustering tendency, non-uniformity. II--Empty Space index.

The here presented Empty Space index (ES) evaluates the fraction of the information space without experimental points, i.e. the space where the distance from an experimental point is significantly larger than the mean distance between the experimental points themselves. ES can be used to eliminate the ambiguity of the some clustering indexes, that aim to evaluate the separation of the data set in clusters, but these clustering indexes are really a mixed measure of clustering, of empty space (the empty space does not necessarily correspond to the break between clusters) and of the degree of uniformity of the objects. The ES index can be used also to correct the MST index, the clustering index based on the distribution of edge lengths in the minimum spanning tree connecting the objects. The corrected MST index seems to be a reliable measure of the clustering degree.

Chemistry Techniques, Analytical↗
PubMedDate not supplied

Zupan's descriptors in QSAR applied to the study of a new class of cardiotonic agents.

Recently a new class of molecular descriptors has been proposed and used in QSAR with simulated data and with regression performed by neural networks. In the present paper these descriptors (Zups, from the name of their author, Juri Zupan) have been slightly modified and then applied to a real data set with the aim of studying the structure-activity relationships of a new class of cardiotonics. Forty-one molecules (thirty-seven milrinone analogues, the two lead compounds amrinone and milrinone, and two commercial products) have been studied using classical chemometrical techniques such as PCA (Principal Components Analysis) and PLS (Partial Least Squares regression). Zups describe essentially the local geometry of the molecules. They show promising performances, as compared with other classical geometrical descriptors (as molecular volume, etc.), both in that regards the overall performances, measured by the C.V. Explained variance and in the interpretability of the regression equation. However they have not all the requirements of a good structure representation. Moreover some selectable parameters seem to have a great importance, so that the refinement of the regression model requires time and the evaluation step must be performed in condition of full-validation, because predictive optimisation is used in the selection of parameters, and the final model must be checked on molecules never used to refine the model or, in this case, the parameters of the structure representation.

Amrinone↗