ML4DB acronymed for Machine Learning for Dittus-Boelter equation, and it is a simple machine learning program (for teaching purpose) written in Python. Dittus-Boelter equation is used to find out the Nusselt Number which depends on Reynolds Number and Prandtl Number based on the type of heat transfer. Exact solution is available for this equation.
- The database is derived from the exact equation and provided in DittusBoelterDatabase.csv. One can regenerate the database with the attached MATLAB script.
- Attached first Python script can be used to read, train and test the DittusBoelterDatabase.csv. It shows the uses of different machine learning algorithm with some initial guess of hyperparameters.
- Second script shows the implementation of deep neural network (DNN) on the same database. Grid search, which is often used to find best hyperparameter, is also included.
- Third script shows the implementation of DNN on the same database using TensorFlow. A high level API for TensorFlow, named Keras, is used.
- Python3
- NumPy
- Sklearn
- Matplotlib
- Pandas
- TensorFlow
- Keras
- seaborn