Linear modeling using topological indices

We start our investigation with 30 topological indices contained inWlOLGEN-QSPR (see Appendix A)  

First we reduce the corresponding set of descriptors since there are various dependencies  

Since there are only single bonds and no heteroatoms in decanes, and are identical. Therefore we exclude x X- By definition, the molecular formula 

The remaining 18 descriptors are obtained from the following functions  

Randomization experiments were performed to compare our models with the random results. Thus, 50 random numbers were defined as a pseudo-y vector, as well as 18 vectors of 50 random numbers acting as pseudodescriptors. By searching all subsets, the best fitting combination of n pseudodescriptors was obtained. For a constant n, 100 such runs were performed, each on a fresh set of random numbers. The highest random F from each run was recorded. The mean highest random (mkrR ) and its standard deviation (stdev) were calculated and are as follows  

---