Watson Analytics – Getting Started

Watson Analytics is a separate effort by IBM to bring Watson’s cognitive capabilities to analytics. It basically exposes Tableau like functionality except it is in the cloud and it uses IBM Watson in the back.

There are three subscription levels: Free, Plus and Professional. You can sign up for a free account that gives you 30 days of access to all of Plus and Professional level features. Once you have done that, you can import up to four of any of the sample data samples available. From there, you can follow one of the use cases available on the IBM Watson Analytics’ website.


Once you get a hang of things, and have some of your own data, you can go ahead and upload new data either from a flat csv file or from one of the several other data input sources available via Watson Analytics including twitter.


All of the functionality of Watson Analytics is distributed between the three Ds (Data, Discover and Display) that you can reach by clicking on the tabs available on the home page.


You would use the Data tab to import Watson Analytics sample data, or upload your own data or pull in from various data feeds. This is also where you would review and refine it imported data.

After you are done massaging and refining the data, you would move on over to the Discovery tab and this where most of the magic would happen. This is where the Watson cognitive services part of analytics starts to shine. For starters, you will be presented with a bunch of queries that Watson Analytics thinks would be important to you.


But what you would likely rather do is based on your data, type in some specific questions. For example, you may type in a question saying “How are Monthly charges related to Churn”. Immediately you are presented with some relevant queries. The image next to the presented queries specify what kind of visualization you are going to get.


You can then click on one of the presented queries, or even build your own, to start the visualization.


These visualizations are active and responsive. As you learn more, you can further drill down by filtering or create new visualizations. For example, we wanted to see the breakdown of InternetService and Contract, and filter the results by Churn column identifying the rows where the customers were lost. This led to the following visualization, that leads us to conclude that customers who had Fiber Optic and were on the month-to-month made up the largest section of the customers lost.


Now we may want to look at churn as a whole, so we can remove that filter for churn for lost customers, and just add it as a whole. This way we end up with a view for Lost and Retained Customers, split by Internet Service, and each segment showing customers that were on month-to-month, one year and two year contracts.


You may want to read this article on IBM detailing tips on asking questions and building queries for watson analytics.

Finally you can move on to the last Display tab to create dashboards using your findings.

To learn more, you can follow one of the use cases available on the IBM Watson Analytics’ website.


Watson Analytics – Getting Started