Energy consumption of biofuels has increased over the past few years relative to fossil fuels, and biomass-based ethanol has attracted interest due to its versatility. The production of bioethanol is a complex process in which the hydrolysis of cellulose plays a decisive role in the final ethanol yield. Moreover, the effectiveness of biomass pretreatment is crucial to determine the enzymatic accessibility and hydrolysis capacity of biomass. Current common studies on biomass pretreatment usually set pretreatment parameters as independent variables without understanding the effect on biomass chemical composition. In this study, a Random Forest-based prediction model was developed to predict the enzymatic hydrolysis of cotton fibers for glucose yield based on the chemical composition (cellulose, hemicellulose, and lignin) of pretreated cotton fibers. Enzymatic hydrolysis of poplar wood samples pretreated with water, acid, and alkali was performed, and the fermentable glucose content obtained was determined, constituting a dataset of 33. To find suitable model parameters, the random forest model was tuned using random search, genetic algorithm, particle swarm algorithm, and simulated annealing algorithm. It was found that the model tuned using a simulated annealing algorithm gave the best results with R2 of 0.92 and RMSE of 2.94 in the test set. The model tuned using a random search algorithm gave the second-best results, while the model tuned using a genetic algorithm and particle swarm algorithm showed underfitting. Therefore, the random forest model optimized based on the simulated degradation algorithm is more suitable for predicting the hydrolyzed glucose content of kapok fibers.