Prediction of Kovats Retention Indices for Fragrance and Flavor using Artificial Neural Network
Aga Maulana, Teuku Rizky Noviandy, Rinaldi Idroes, Novi Reandy Sasmita, Rivansyah Suhendra, Irvanizam Irvanizam
Abstract
In this research the Kovats retention index of 51 fragrance and flavor substances was successfully predicted using Artificial Neural Network (ANN). ANN has been run three times with each number of iterations of 1000, 5000, and 10000. These three iterations are chosen to see the best iteration in generating the R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> and RMSE. This research indicates that the number of iterations of 5000 is the best iteration after testing. The study obtained R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> = 0.986 and RMSE = 28.99, with an average difference between predicted and observed is 2.5%. From these results, it can be understood that the ANN model can predict the Kovats retention indices of fragrance and flavor substance quite well.