Optimization of EDTA enriched phytoaccumulation of zinc by Ophiopogon japonicus: Comparison of Response Surface, Artificial Neural Network and Random Forest models

Abstract Zinc (Zn) contaminated soil was remediated by EDTA enriched phytoaccumulation strategy using Ophiopogon japonicus . The main and interactive effects of three variables such as initial Zn concentration (0.5 × 10 −4  M–1 × 10 −4  M), EDTA concentration (0–25 mmol kg −1 ), and time period (1–14 days) were investigated via Box–Behnken statistical design (BBD). The optimal phytoaccumulation efficiency of 95.30% was found to be at Zn concentration: 1 × 10 −4  M kg −1 , EDTA concentration: 16.42 mM kg −1 , and time period: 13.86 days. The same data was given as input to Multilayer Feed-Forward Networks Back-Propagation and Random Forest (RF) model to classify the data as high and low yield in order to develop a real-time monitoring system. The performance of the aforementioned models for Zn phytoaccumulation was evaluated in terms of parity plot, normal probability plot, and error functions and RF model was found suitable.