Watson, is That You? Yes, and I’ve Just Demo-ed My Analytics Skills at IBM’s New York Office


My photo of the entrance to IBM’s office at 590 Madison Avenue in New York, taken on July 29, 2015.

I don’t know if my heart can take this much excitement. Yesterday morning, on July 29, 2015, I attended a very compelling presentation and demo of IBM’s Watson technology. (This AI-driven platform has been previously covered in these five Subway Fold posts.) Just the night before, I saw I saw a demo of some ultra-cool new augmented reality systems.

These experiences combined to make me think of the evocative line from Supernaut by Black Sabbath with Ozzie belting out “I’ve seen the future and I’ve left it behind”. (Incidentally, this prehistoric metal classic also has, IMHO, one of the most infectious guitar riffs with near warp speed shredding ever recorded.)

Yesterday’s demo of Watson Analytics, one key component among several on the platform, was held at IBM’s office in the heart of midtown Manhattan at 590 Madison Avenue and 57th Street. The company very graciously put this on for free. All three IBM employees who spoke were outstanding in their mastery of the technology, enthusiasm for its capabilities, and informative Q&A interactions with the audience. Massive kudos to everyone involved at the company in making this happen. Thanks, too, for all of attendees who asked such excellent questions.

Here is my summary of the event:

Part 1: What is Watson Analytics?

The first two speakers began with a fundamental truth about all organizations today: They have significant quantities of data that are driving all operations. However, a bottleneck often occurs when business users understand this but do not have the technical skills to fully leverage it while, correspondingly, IT workers do not always understand the business context of the data. As a result, business users have avenues they can explore but not the best or most timely means to do so.

This is where Watson can be introduced because it can make these business users self-sufficient with an accessible, extensible and easier to use analytics platform. It is, as one the speakers said “self-service analytics in the cloud”. Thus, Watson’s constituents can be seen as follows:

  • “What” is how to discover and define business problems.
  • “Why” is to understand the existence and nature of these problems.
  • “How” is to share this process in order to affect change.

However, Watson is specifically not intended to be a replacement for IT in any way.

Also, one of Watson’s key capabilities is enabling users to pursue their questions by using a natural language dialog. This involves querying Watson with questions posed in ordinary spoken terms.

Part 2: A Real World Demo Using Airline Customer Data

Taken directly from the world of commerce, the IBM speakers presented a demo of Watson Analytics’ capabilities by using a hypothetical situation in the airline industry. This involved a business analyst in the marketing department for an airline who was given a compilation of market data prepared by a third-party vendor. The business analyst was then assigned by his manager with researching and planning how to reduce customer churn.

Next, by enlisting Watson Analytics for this project, the two central issues became how the data could be:

  • Better understand, leveraged and applied to increase customers’ positive opinions while simultaneously decreasing the defections to the airline’s competitors.
  • Comprehensively modeled in order to understand the elements of the customer base’s satisfaction, or lack thereof, with the airline’s services.

The speakers then put Watson Analytics through its paces up on large screens for the audience to observe and ask questions. The goal of this was to demonstrate how the business analyst could query Watson Analytics and, in turn, the system would provide alternative paths to explore the data in search of viable solutions.

Included among the variables that were dexterously tested and spun into enlightening interactive visualizations were:

  • Satisfaction levels by other peer airlines and the hypothetical Watson customer airline
  • Why customers are, and are not, satisfied with their travel experience
  • Airline “status” segments such as “platinum” level flyers who pay a premium for additional select services
  • Types of travel including for business and vacation
  • Other customer demographic points

This results of this exercise as they appeared onscreen showed how Watson could, with its unique architecture and tool set:

  • Generate “guided suggestions” using natural language dialogs
  • Identify and test all manner of connections among the population of data
  • Use predictive analytics to make business forecasts¹
  • Calculate a “data quality score” to assess the quality of the data upon which business decisions are based
  • Map out a wide variety of data dashboards and reports to view and continually test the data in an effort to “tell a story”
  • Integrate an extensible set of analytical and graphics tools to sift through large data sets from relevant Twitter streams²

Part 3: The Development Roadmap

The third and final IBM speaker outlined the following paths for Watson Analytics that are currently in beta stage development:

  • User engagement developers are working on an updated visual engine, increased connectivity and capabilities for mobile devices, and social media commentary.
  • Collaboration developers are working on accommodating work groups and administrators, and dashboards that can be filtered and distributed.
  • Data connector developers are working on new data linkages, improving the quality and shape of connections, and increasing the degrees of confidence in predictions. For example, a connection to weather data is underway that would be very helpful to the airline (among other industries), in the above hypothetical.
  • New analytics developers are working on new functionality for business forecasting, time series analyses, optimization, and social media analytics.

Everyone in the audience, judging by the numerous informal conversations that quickly formed in the follow-up networking session, left with much to consider about the potential applications of this technology.

1.  Please see these six Subway Fold posts covering predictive analytics in other markets.

2.  Please see these ten Subway Fold posts for a variety of other applications of Twitter analytics.


2 thoughts on “Watson, is That You? Yes, and I’ve Just Demo-ed My Analytics Skills at IBM’s New York Office

  1. Pingback: New Startup’s Legal Research App is Driven by Watson’s AI Technology |

  2. Thank you for sharing all those details about Watson. It is exciting news that IBM is equipping Watson for medical diagnosis. Clinical decision making systems are one of the key areas of health information technologies, and artificial intelligent systems like Watson have advantages in this area.

    Firstly, making a medical diagnosis requires a large database. Traditionally this database resides in doctors’ memories. Doctors connect the observed symptoms with possible health conditions and deliver the best match as the diagnosis. This process may be simplified into two steps: (a) remembering diseases and their matching sets of symptoms, also referred to as training; and (b) matching observed symptoms with sets of symptoms in memory, also called testing. For machines, these steps may take less time and effort than humans. This is because theoretically they can have limitless memories and high computation speed.

    Lastly, Watson can process natural languages. This frees doctors’ hands so they can focus on issues such as emergency treatments while talking to Watson at the same time. When lives are on the line, the time that doctors save from having to enter commands and search terms can be of critical importance.

    These characteristics can potentially help patients in their home. Let us consider type 1 diabetes for an example.

    First, Watson can help with decision making at home. People living with type 1 diabetes usually need multiple insulin injections a day and different health status throughout the day may require different dosages. The health status, however, can be influences by a number of factors ranging from blood sugar levels, diet, exercise and stress to age, weather and seasons. While it is hard for humans to account for every factor when determining insulin dosages on a daily basis, Watson can finish the personalized computation in a split second. This ability may be especially desirable to newly diagnosed diabetes patients because they are unfamiliar with how their bodies react to insulin, their behaviors and the environment.

    Second, Watson’s conversation ability can reduce the stress of patients trying to record everything. Patients can talk to Watson to create new records, making the recording process more efficient.

    Third, as a machine, Watson has the potential to connect to patients’ existing devices, such as the insulin pump, continuous glucose monitor and smart phone apps. Using this data, Watson can make devices talk to each other and potentially closing the loop for health status data collection and insulin administering. Although recent developments in technologies for diabetes patients have accomplished this, Watson can still be more valuable by taking into consideration factors other than blood sugar levels and diet.

    However, Watson can face challenges when applied in the home and on some medical processes other than diagnosis.

    First, size matters. A Watson terminal can be much bigger than the modern insulin pumps that we are used to. It would be almost impossible to carry it around. If we host Watson on a super computer and use mobile technologies to access its intelligence, then we are at the mercy of the availability and stability of wireless connections. This situation, however, can be greatly improved if we can only carry a portion of Watson’s intelligence in a small device.

    Second, effectively using Watson may cause analysis paralysis. Many patients I talked to have an intuition of how much insulin to administer and they reported that over analyzing can cause over compensation, leading to unwanted high or low blood sugars. When we try to take advantage of the advanced computing power offered by Watson, we may be over thinking what we are doing every day and experience unnecessary anxiety.

    Third, Watson has superior problem solving and decision making abilities (using information collected in the past) but may be less successful in planning (anticipating the future) and persuading (understanding emotions). Human factors have always been a challenge in computer engineering. We have started to build social robots that can offer companionship, but the field is still in its infancy, and is very worth looking forward to.

    Watson is picking a unique route toward medical information technology advancement. Natural languages are slowly moving out of the clinical environment because of its error prone and time consuming nature. Watson addresses these two issues gracefully. Perhaps in the future, Watson would also overcome the difficulties with application in the home and become our personal health assistant.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.