A quantitative evaluation of explainable AI methods using the depth of decision tree

It is necessary to develop an explainable model clarify how and why a medical makes particular decision. Local posthoc AI (XAI) techniques, such as SHAP LIME, interpret classification system predictions by displaying the most important features rules underlying any prediction locally. Therefore, in order compare two or more XAI methods, they must first be evaluated qualitatively quantitatively. This paper proposes quantitative evaluation metrics that are not based on biased subjective human judgment. On other hand, it dependent depth of decision tree (DT) automatically effectively measure complexity methods. Our study introduces novel strategy measures method using characteristic another proxy. The output specifically feature importance scores from fed into DT our proposal. will then draw full score decisions. As result, we developed main can used assess DT?s thus associated method: total (TDT) average weighted class (ACD). results show outperforms LIME less complex. Furthermore, terms number documents features, scalable. These indicate whether specific suitable for dealing with different document scales. demonstrate which improve performance black-box model, this case, feedforward neural network (FNN).