Abstract
This paper demonstrates the operability and effectiveness of a Stacking ensembled classification model applied to infrared spectral analysis for identifying the origin of Chinese herbal medicines. The study focuses on dimensionality reduction based on feature importance weights through the embedding method, which proves to be the most stable and effective approach in this context. The results show that the ensembled Stacking classification model outperforms single classification models, leading to accurate and effective predictions in the test dataset.