Models and datasets for "Learning effective good variables from physical data"

Trained models and datasets related to the GitHub repository https://github.com/giuliobarl/GoodPhysVariables and to our publication (https://www.mdpi.com/2504-4990/6/3/77)

• Dittus.xlsx, Gnielinski.xlsx, and Newton.xlsx are the original datasets with no noise;
• Dittus Noise.xlsx, Gnielinski Noise.xlsx, and Newton Noise.xlsx are the datasets to which gaussian noise has been added, and that are used to train the DNN models;
• dittus_model.tf.zip contains the weights of the trained DNN model for the Dittus-Boelter correlation;
• gnielinski_model.tf.zip contains the weights of the trained DNN model for the Gnielinski correlation;
• newton_model.tf.zip contains the weights of the trained DNN model for Newton's law of universal gravitation.