TensorFlow ML Regression Script Creation

1. Create a script to train on the large dataset with the following implementation requirements:
a. this is regression task
b. use tf.data.Dataset for data pipeline (tf.data.experimental.CsvDataset is acceptable)
c. have a preprocess function that batch-normalizes INPUT features (this is NOT the same as batch normalizing of model weights) when mapped on to the dataset object
d. placeholder for custom loss function with RMSE implementation
e. model must be coded in OOP style (NOT using Sequential or Functional style)
f. model must be saved at end of each training cycle, and can be reloaded for further training (with more epochs)
g. there should be 2 placeholder classes for 2 models
i. a simple ANN (with dense layers and LeakyReLU activation function)
ii. a LSTM model for assuming this is a multivariable time series regression problem)

2. Create a 2nd script to load the trained model above and evaluate it on a directory of test CSVs
a. also use tf.data.Dataset
b. propose how you would handle the preprocessing step for the test data


3. Document code in NumPy style (https://numpydoc.readthedocs.io/en/latest/format.html)