Outside the physical sciences, few systems or properties can be described precisely in terms of a single variable. The major part of scientific and industrial investigations results in data that are multivariate, and, in addition, collinear.

Approximately in all of variable selection methods have been used from dependent variables and used for a specific case, so we can called them as a supervised variable selection. If there is a way to select the informative variable without considering to the dependent variables, we can say that selected variables have important and informative part of variance in data and can be used for any set of dependent variables. Some methods such as PCA and PLS reduce data variables to the much less than variables (Latent Variables), which do not have physical meaning.

Here, we want to investigate a method to select the informative part of a data without considering to the dependent variables, also from original variables.

At first Gram-Schmidt Orthogonalization (GSO) were used in order to removing collinearity and redundancy between the descriptors. Selection of informative variables is based on PLS modeling of orthogonalized data with principal components of independent variables. Then, Jack-knife resampling with a significant test applied as a criterion for selection of informative variables.

1. Supramolecular chemistry in DNA (swf)

2. 5th International Conference and workshop on Mathematical Chemistry - Yazd - Iran Jack-knife based variable selection without dependent variables (swf)

3. Improved variable reduction in Partial Least Score modelling based on Predictive-Property-Ranked variables and adaptation of partial least squares complexity (swf)

4. Is your QSAR/QSPR descriptor real or trash? (swf)

5. Advances in the Replacement and Enhanced Replacement Method in QSAR and QSPR Theories (swf)

6. Variable selection in near infrared spectroscopy based on significance testing in partial least squares regression (swf)

7. Evaluation of rotation ambiguities in Multivariate Curve Resolution (swf)

8. Resolution of Ambiguities in Structure-Property Studies by Use of Orthogonal Descriptors (swf)

9. Multivariate Analysis (swf)

10. Successive Projections Algorithm (swf)