Abstract:

Modern data mining employs the use of statistics with other tools, ideas and approaches from computer science, database technology, machine learning, deep learning as well as other classical analytical tools. In this research, three distinct algorithms have been employed for modelling the performance properties of Diazepam (DIA) utilizing the method of high-performance liquid chromatography (HPLC). The algorithms developed include support vector machine (SVM), adaptive-neuro fuzzy inference systems (ANFIS), and linear regression (LR). Temperature and mobile phase inform of methanol (MeOH) were used as predictors, while the time recorded for the retention was deemed to be the dependent variable. The performance accuracy of the models was assessed using two statistical metrics, including determination co-efficient (R2) and root mean square error (RMSE). The obtained results were shown both qualitatively and graphically using different charts. The comparative performance accuracy of the models demonstrates that the non-linear models (ANFIS and SVM) displays a higher performance efficiency than LR and has the ability of enhancing its performance ability by 51.2% and 76.1% both throughout the training and testing phase.