Quality monitoring during sufu fermented process and establishment of random forest prediction model

Abstract

SummaryMicroscopic, mechanical, physiochemical and color changes of sufu during fermentation (10d, 20d, 30d, 40d, 50d, 60d) were analyzed to predict sufu quality objectively and accurately. Fourier transform infrared spectroscopy (FT‐IR) and scanning electron microscopy (SEM) showed that the absorption peaks at 702 cm‐1, 1454 cm‐1, 1652 cm‐1, 2926 cm‐1 and 3281 cm‐1 increased in intensity, and sufu was altered from a granular colloidal protein structure to a smooth and uniform shape. Analysis of mechanical, physiochemical and color properties demonstrated the growth of mucor, the addition of salt and dressing mixture had a significant impact on the changes of indexes of sufu with extension of fermentation time. Besides, a random forest (RF) sufu quality prediction model with 90% accuracy was established by selecting forest quality evaluation indicators using RF and setting the model parameters mtry and ntree to 5 and 900, respectively.