Implementation based on Yellowbrick library adapted to directly support trained Deep Learning models from TensorFlow and Keras.
Currently supports visualizations for Classification Report, Confusion Matrix, ROC-AUC and Class Prediction Error.
classes = ['Cat', 'Dog']
name = 'VGG19'
path_to_save = './images'
if not os.path.exists(path_to_save):
os.makedirs(path_to_save)
vz = MetricsVisualizer(model, X_test, y_test, classes, name, path_to_save)
vz.ClassificationReportViz()
vz.ConfusionMatrixViz()
vz.ROCAUCViz()
vz.ClassPredictionErrorViz()