We can increase the flexibility of our training experiment, by adding parameters to our script which will allow us to repeat the same Training experiment with different settings. So in this case, we'll add a parameter for the regularization rates used by the logistic regression, of when we're training the model.

So again, let's start by creating a folder for the parameterized script and the training data. So we need an os model and our shutil model. As before we create a folder for the experiment and then using shutil, we copy the diabetes CSV to the folder.

So now let's create a script considering the parameter for the realization rate, a hyper barometer. So using a magic command, right file. You specify the folder and the script we're writing into. We need to import the following libraries. So we need run. We need our pandas. We need numpy and we need joblib. We need os. And we need argparse, which allows us to pass in arguments into a script from the command line. We need a train test split. We need our logistic regression algorithm. We need our metrics, here under the curve and accuracy.

With our imports done, let's get the experiment context. So to set our realization, hyper parameter, we invoked ArgumentParser off argparse. That's created a parser object. We invoke add argument. We pass in the name, type, and then the name of the attribute and a default value, but then load the diabetes data set within sep, separate the features and labels, split the training set, and test sets. And then we train our logistic regression model and log the realization rates.

After training, we can make a prediction and then we can measure the accuracy of our prediction and load the information using our run and log function within predict probability. And we can use that information to calculate our area under the curve and also store that information using our log functional run object. And finally, we can store our model and then complete our run.