<p>Co-pyrolysis of biomass with plastics can significantly increase the added value of biochar. Machine learning techniques can effectively predict the yield of biochar from biomass and plastic co-pyrolysis. This study innovatively introduces data from the co-pyrolysis of the three major biomass components (cellulose, hemicellulose, and lignin) with plastics, including both individual and mixed biomass components, improving upon the limitation of traditional research that focuses solely on the overall characteristics of biomass. Based on the Light Gradient Boosting Machine (Light GBM) and Deep Neural Network (DNN) algorithm, a predictive model for biochar yield from biomass and plastic co-pyrolysis was developed. The results show that LightGBM outperforms DNN. The introduction of these data significantly improves LightGBM’s prediction accuracy and performance, with an average increase in R² of 0.07 and an average decrease in MAE and RMSE by approximately 2. LightGBM_c achieves the best performance, with an R² of 0.901, MAE of 3.672, and RMSE of 4.792. This study proposes a novel approach to improve predictive models in thermochemical conversion research involving biomass and plastics, which provides valuable insights for optimizing the preparation of biochar and increasing biochar yield.</p>