Chinese herbal medicine classification prediction using random forests under the constraint of infrared spectral data

The same herbs from various origins were scanned with mid-infrared under standard lighting conditions, and the spectral readings at various wavelengths were gathered, sorted, examined, and downscaled. The downscaled data were then divided into a training set and a test set to form modeling data, which were used to determine known herb origins and predict unknown herb origins using a random forest model in Python and contrasted with a linear support vect. The paper notes that, when comparing the two models, random forest and support vector machine, for herbs of known origin, the random forest model performs better in terms of dimensionality reduction, correct recognition rate in training sets, and correct recognition rate in test sets, by 28.57%, 1%, and 2.9% respectively. The findings demonstrate that the random forest model is deserving of being of significant value in the era of big data to assist clinicians in precisely identifying herbal medicine categories for treatment. It is effective in boosting classification accuracy and lowering computation and time consumption.