<![CDATA[AbstractThere are two methods of approach in solving computational chemistry problems which are computational quantum chemistry and computational non-quantum chemistry. Computational quantum chemistry methods are e.g. ab initio and semiempirical, whilst the non-computational quantum chemistry method is molecular mechanics. This study was aimed to find the difference ofcomputational quantum chemistry and non-computational quantum chemistry methods by applying ab initio, semiempirical and molecular mechanics methods to predict the physicochemical properties (Log P, melting point) and spectrum (ultravioletvisible, 1H-NMR, 13C-NMR) of quercetin, glucosamine and andrographolide as a representation of natural bioactive compounds. Results showed that the minimum energy of geometry optimization provided by ab initio methods, while the fastest time for geometry optimization given by molecular mechanics method. Log P prediction was not influenced by the geometry optimization, with error rate: MAD 0.271; MSE 0.204; MFE 0.255 and MAPE 9.66%, meaning the accuracy was good. Melting point prediction was not influenced by the geometry optimization, with error rate: MAD 190.96; MSE 55434.78; MFE -190.96and MAPE 70.82%, meaning the accuracy was not good. Ultraviolet-visible spectrum prediction was affected by the geometry optimization. Ab initio method was the best for predicted λ maximum with error rate: MAD 2.78; MSE 7.90; MFE 2.78 and MAPE 1.15%. The error rate of semiempirical method: MAD 6.46; MSE 42.06; MFE 6.46 and MAPE 2.73%. The error rate of molecularmechanics method: MAD 30.93; MSE 958.65; MFE 30.93 and MAPE 12.94%. The results of 1H-NMR and 13C-NMR spectra prediction was not influenced by the geometry optimization and chemical shift in accordance with the theory of NMR spectra.Keywords: ab initio, semiempirical, molecular mechanics, quercetin, glucosamine, andrographolide]]>
