This Repo is used for providing fellow students with a deep learning intro.
A Artificial Neural Network is built from ground up (except for the gradient calculations which are done with tf.GradientTape()) and trained on the titanic dataset, reaching an acurracy of about 80%.
Different metrics like loss, accuracy, true_pos etc that are calculated during training get plotted. After training a small command line interface allows to enter features of an imaginary person to find out if they would have survived the accident.