Comparisons between the performance of the model retrained after excluding a certain feature and the performance of the model before exclusion. The red bars and purple bars represent the percentage change in mean accuracy and mean logloss during the 5-fold cross-validation process respectively. As can be seen, stellar mass, M_*, and gas mass, M_g, have the most substantial impact on the overall performance of the classifier, indicating their crucial importance in MLCCA inference.