After training the baseline model, evaluate its performance using a confusion matrix. This will provide an overview of classification results across different categories.
Perform error analysis by extracting the misclassified samples. Pick one pair of frequently confused pair of genre and one pair of completely opposite types of genre. Compare their average feature values and visualize the differences with a bar plot or heatmap to illustrate how feature similarities/differences influence misclassification.
After training the baseline model, evaluate its performance using a confusion matrix. This will provide an overview of classification results across different categories.
Perform error analysis by extracting the misclassified samples. Pick one pair of frequently confused pair of genre and one pair of completely opposite types of genre. Compare their average feature values and visualize the differences with a bar plot or heatmap to illustrate how feature similarities/differences influence misclassification.