Remember that you can recuse yourself from doing what this section is doing and simply pay close attention to it if you are afraid to be charged. Nonetheless, within the free trial, I was able to do the following, and even if you were charged full price, it's not likely to be expensive.
Visit the following link for more details about Azure Machine Learning Studio pricing: https://azure.microsoft.com/en-us/pricing/details/machine-learning-studio/.
You can check which Free Trial is on by logging into https://account.azure.com and then clicking ACCOUNT and subscriptions from the top menu. The following screenshot is highlighting the path:
As you can see, free trial was still on with 24 days to go when I checked. The red circles show where you have to click to check it for yourself: ACCOUNT, subscriptions. Right next to SIGN OUT in the top-right corner, there is a blank space, and that is where the email is displayed.
Some billing information may be displayed if there is anything to be billed. Another way to check the billing information is by consulting the Azure portal. Access the portal through the following link: https://portal.azure.com.
After logging in, search for the topic Cost Management + Billing on the left menu, as shown in the following screenshot:
Azure portal displays this cool dashboard that you can modify at will—customization here is very cool. By clicking the icon circled in red (Figure 13.2), the user lands on a page with more detailed billing information. The following screenshot gives an example of what can be found in it:
I was still in the free trial, so there wasn't anything to pay. Azure Machine Learning Studio can also be accessed through the Azure portal, which requires some extra steps when compared to directly accessing it through the following link: https://studio.azureml.net.
The home page may look like the following screenshot:
In the top-right corner, there is a Sign In button, click on it to reach for the platform. If you are already logged into Azure, you will be automatically taken to Azure Machine Learning Studio interface. If you are not, log in using your Azure credentials, and you will be led to the Machine Learning Studio interface.
As soon you log into http://studio.azureml.net, you might see something such as this:
By choosing Take Tour, the user will be guided to a brief quick-start tutorial. It will take around five minutes to complete. Take a break from reading and join the tour. You won't be introduced to R in Studio, but it will present the very same tool that this chapter will trust to deploy R code using Machine Learning Studio.
In the bottom-left corner, search for a plus sign next to the word NEW, written in uppercase, and select it. The following screenshot shows where to find the button:
Clicking the singled-out icon will unleash a whole new blade with lots of options. A user can choose either to create a new DATASET, MODULE, PROJECT, EXPERIMENT, or NOTEBOOK. There is also an X button near the upper-right corner; click it to go back. This whole new blade is illustrated in the following screenshot:
As Figure 13.7 demonstrates, to begin a new experiment, a user has to pick an experiment template. I am going with Blank Experiment, but, before you go on and select the same, take a quick detour and look at—only look—which other templates are available.
After inaugurating Blank Experiment, you will be given a taste of the cool diagram-chart format that comes with the Azure Machine Learning Studio Experiments. Check the following screenshot for an example:
Why craft wonderful visualizations for others to see and not have some cool stuff aiding us during work time? The Experiment interface helps data scientists to visualize the whole landscape of their work while not missing the connections between individual steps.
On the left, there is a white blade with several topics: Saved Datasets, Trained Models, Data Format Conversions, and so on; there is even a topic for R language modules. Under each of these topics, there are one or more modules. Experiments are built by dragging, dropping, and connecting modules into the middle area.
Each module will handle one or more tasks. Some examples of tasks handled by individual modules are the following: loading data, converting data to a different type, splitting DataFrame, training models, making predictions, scoring, and evaluating models. By dropping and combining individual modules, complete data science models can be delivered.
On the right side (Figure 13.8), there is another blade. What you can do with it depends whether an active module is selected. When none is selected, a user can add descriptions about the experiment, which is a good way to document it. When a module is selected, users can change its options (parameters) if it has any.
Azure allows collaboration, and you may also want to visit your experiment later. For the sake of organization and understandability, do your best to keep objective and concise notes and comments—it pays well to do so in the long run. A particular worthwhile thing to remember with respect to Studio's Experiments is to know how modules work in general.