Abstract
The intention of this present study is to optimize the recovery of pectin from Persea americana peel (PAP) examining four independent process variables (pH, solvent to substrate ratio (SSR) (ml/g), agitation time (AGT) (h), and agitation speed (AGS) (rpm)) and compare the pectin recovery (PR) using statistical (Box-Behnken response surface design (BBRESD)) and non-statistical (artificial neural network (ANN) with genetic algorithm (GA)) methods. Optimal condition derived in ANN-GA (pH of 1.9, SSR of 16 ml/g, AGT of 2.1 h, AGS of 99 rpm, and PR of 90.59%) has forecast PR very precisely than BBRESD (pH of 2, SSR of 14 ml/g, AGT of 2 h, AGS of 100 rpm, and PR of 85.44%). Extrapolative capacities of two methods were examined by several statistical restraints. Structural analysis (FT-IR and XRD) of PR was exhibited that highly esterified and amorphous nature of pectin was recovered from PAP. Recovered pectin was melted at the temperature of 258.5 degrees C which was similar to commercial pectin. SEM observations revealed that recovered pectin was flake and spherical particle. The results of this work suggested that pectin from PAP is one of the good sources to recover pectin with good quality as commercial pectin and used as a food ingredient.