Use of Decision Tree Algorithm to Model 2,4-D Herbicide Removal Process from Water by Rice Husk Biochar

This study modeled the adsorption process of 2,4-D pollutant from the aqueous medium by rice husk biochar adsorbent using CART and CHAID decision tree algorithms. The effect of contact time (0-120 min), temperature (20-60 oC), initial concentration of contaminant (60-600 mg/L), adsorbent dose (0.05-0.2 g), and pH (2-9) was examined on the adsorption capacity. The results indicated that the adsorption capacity reached the maximum at pH=5.50 and the temperature lower than 30 oC. The outputs of both models showed that the initial concentration of the contaminant and the adsorbent dose had the most significant effect on the adsorption capacity, increasing with the initial concentration increment and the adsorbent dose decrement. At the initial concentration of less than 350 mg/L and the adsorbent dose of less than 0.08 g, the maximum adsorption capacity was 37.1 mg/g. The CART model, with an R2 of 98.1%, showed higher performance than the CHAID model, with a 97.7% value, in estimating the adsorption capacity of rice husk biochar. Rice husk biochar (BRH) can be used as a suitable adsorbent to remove 2,4-D herbicide from water due to its low initial cost, high physical and chemical capability, and high ability to remove the studied contaminant.