Computer Vision News - April 2020

3 Summary A Step Towards Explainability 17 Would that make sense? Well yes, as more blood arrives at the heart, so it is indeed expected, as shown on the PDP. Remember the PDP plots with two variables? Let’s use them to examine the relation on ‘ST depression induced by exercise relative to rest’. inter1 = pdp.pdp_interact(model=model, dataset=X_test, model_features=base_features, features=['st_slope_upsloping', 'st_depression'])pdp.pdp_interact_plot(pdp_interact_out=inter1, feature_names=['st_slope_upsloping', 'st_depression'], plot_type='contour')plt.show()inter1 = pdp.pdp_interact(model=model, dataset=X_test, model_features=base_features, features=['st_slope_flat', 'st_depression'])pdp.pdp_interact_plot(pdp_interact_out=inter1, feature_names=['st_slope_flat', 'st_depression'], plot_type='contour')plt.show() In both cases, it seems that low depression is negatively correlated. On the paper of G.A. Lanza et al. it is noted that assessing ST-segment depression during recovery can significantly improve the clinical information derived from exercise stress tests, thus this isn’t the expected result and we can examine it further by looking at the SHAP values. explainer = shap.TreeExplainer(model) shap_values = explainer.shap_values(X_test) shap.summary_plot(shap_values[ 1 ], X_test, plot_type="bar")