Different Aspects in Partial Least Squares and Artificial Neural Network Models Used for the Analysis of Cefoperazone Sodium in Presence of its Alkaline Degradation Product

Abstract

Several chemometric models were used for the determination of cefoperazone sodium in presence of its alkaline degradation product. The methods are either traditional (Partial Least Squares) or advanced (Artificial Neural Network). Partial Least Squares method was used with and without variable selection (Genetic Algorithm GA). Artificial Neural Network (ANN) was used with and without variable selection procedure (Genetic Algorithm GA) and data compression procedure (principal component analysis PCA). The chemometric methods used are PLS-1, GA-PLS-1, ANN, GA-ANN and PCA-ANN. The methods were used for the determination of cefoperazone sodium in bulk powder and pharmaceutical preparation. A 2-factor 5-level experimental design was built leading to 25 mixtures containing different ratios of cefoperazone sodium and its alkaline degradation product. Thirteen mixtures were used as a training set and the other twelve were used as a validation set.