Watson Analytics

Need for Watson Analytics
If an organization is good at analyzing data and extracting relevant insights from it then decision makers can make more informed and thus more optimal decisions. But the decision makers are forced to make decisions with incomplete information. The reason?  Decisions makers/ Citizen Analysts, for the most part, tend to be mainly consumers of analytics and they rely on more skilled resources (Like Data Engineer, Data Scientist, Application developer) in the organization to provide the data driven answers to their questions. Moreover the answer to one question is just the start of another. Think of a detective interrogating a suspect. The consumer/builder model is hardly conducive to the iterative nature of data analysis. Therefore, the time it takes for these answers to be delivered to the decision makers is far from optimal – and many questions go unanswered every day.

watsonlogoWatson Analytics
So a logical solution is to provide an easier to use analytics offerings. Watson Analytics provides that value add so that more people will be able to leverage data to drive better decision making using analytics.

When we think of Watson, we think about Cognitive. And when we think about Analytics, we think  about traditional analytics (querying, dashboarding), along with some more advanced analytic capabilities (data mining, and social media analytics). So Watson Analytics is a Cloud based offering which can make analytics a child’s play even for a non-skilled user.

Watson Analytics helps users understand their data in a guided way using a natural language interface to ask a series of business questions. Example, a user can ask “What is the trend of revenue over years?” and get a visualization in response. So, Instead of having to first choose a visualization and working backwards to try answer the business question, Watson Analytics allows you to describe your intent in natural language, and it chooses the best visualization for you. Even better, Watson Analytics gives you some initial set of questions which you can keep refining.

Watson Analytics for Social Media
Watson Analytics can work on Social Media data to take the pulse of an audience by spotting trends and identifying new insights and relationships across multiple social channels allowing greater visibility into a given topic or market. It combines structured and unstructured self-service analysis to enrich your social media analytics experience for exceptionally insightful discoveries. All on the cloud!

Summary of Steps:
Watson Analytics does the following to provide insights hidden in your Big data. Mouse-over the below images to get the details of the steps.

  • Import data from a robust set of data source (on Cloud and on premise) options, with the option to prepare and cleanse via IBM Bluemix Data Connect.
  • Answering What: Identifying issues, early problem detection, finding anomalies or exceptions, challenging conventional wisdom or the status quo.
  • Understanding or explaining outcomes, Why something happened.
  • Dashboarding to share results

Providing relevant Social Media Analytics

Slowly social media is becoming very prevalent place where individuals or groups of individuals express their opinions on the gamut of tools/ services/ latest gizmos/ ad campaigns to an organization as a whole.  An organization should be aware of what the consumers/ customers or competitors are talking about them to decide the future course of plan and remain competitive in market place.  For example on a launch of a new product a negative chatter starts about it. It could have been a misinformation, but this chatter has potential to do the damage to the sales of the product and many consumers may form their opinions. So it would be nice for an organization to scan the social media and get the answers to the questions like following:

  • How do consumer feel about our new product or ad campaign?
  • What are consumer hearing about our brand [Brand reputation]?
  • What are the most talked about product attributes in my product category [Like in my smart phone whether people are talking about screen, battery life or camera]. Is it good or bad?
  • What is my competitor doing to excite market [Competitive analysis]?
  • Are my business partners helping or hurting my reputation?
  • Is there a negative chatter that my PR team should respond to?

Cognos Cosumer Insight (CCI) does the same thing. Typically it does not crawl the data and uses the service(s) of some known social media crawlers (like Board Reader) to get social media content. This ASCII data comes in Jason format and CCI processes the data using Hadoop from BigInsights. It applies sentiment analysis (using SystemT) and does the following

  • Perform pattern matching, based on the input keywords, and then look for sentiments
  • Check for positive, negative, neutral words / phrases (grammar, slang, typos, synonyms etc)
  • Save results for further processing by visualization / search engine.

Finally it provides Sentiment Analysis using the Visualization Engine.  The job of CCI is done now, but the story does not end here. Now we know the sentiment of the customer as of now. Organizations may need the ability understand the key factor driving the sentiment and IBM SPSS is a tool for that. Once there is a fair prediction, we need to act on those and monitor results and IBM Coremetrics and Unica are the tools used for the same.