Readme

Dataset can be found here : https://www.kaggle.com/datasets/pablomgomez21/drugs-a-b-c-x-y-for-decision-trees
Instead of using decision trees for this dataset, this solution shows what can neural network of softmax multiclass classification algorithm can do.
Accuracy results of activation functions :

  • relu : loss: 0.0729 - accuracy: 0.9750
  • swish : loss: 0.0496 - accuracy: 1.0000
  • gauss : loss: 0.0410 - accuracy: 1.0000
  • tanh : loss: 0.0246 - accuracy: 1.0000
  • selu : loss: 0.0242 - accuracy: 1.0000
  • softplus : loss: 0.0151 - accuracy: 1.0000