I have used many tools over the past decade as a data scientist. Yet, Watson Analytics is the one that truly has the ability to democratize analytics.
Why do we need to democratize analytics?
The simple answer is that if we continue to think of it as rocket science and place it on a pedestal, it delays the full benefit to the business. Companies tend to approach analytics with trepidation. There’s a natural tendency to shy away from anything to do with “statistics” or “mathematics.” As a result, many of us avoid taking quantitative analysis head-on. Instead, we leave it to focused analytics teams. Waiting for the teams to provide reports and analysis can slow decision-making to a crawl.
Always room for a data scientist or two or 10…
There will always be room for skilled data scientists who develop highly customized algorithms (trust me; I am one of those guys). For example, consider the three types of analytics. Descriptive analytics identifies key relationships between variables. Predictive analytics predicts target variables based on identified key variables. Prescriptive analytics prescribes optimal levels at which to operate these variables to optimize a performance criterion.
Much of predictive and prescriptive analytics require the skills of trained data scientists, analytics experts and analytics programmers – and probably always will. But, we should all root for tools that democratize the descriptive analytics process for all business roles in an organization. Moreover, if there are tools that can help them ease into basic predictive analytics, even better. It’s for the greater good. Smarter decision-making for all, and by all, can in no way be a bad thing, can it?
Watson Analytics serves as a critical component of any business’s analytics journey, especially in the descriptive and predictive phases.
Watson Analytics empowers everyone with basic analytics capabilities necessary to analyze and get new insights from data. Watson Analytics is a guided, autonomous and independent analytics service that shares some genetic material with IBM Watson (of Jeopardy! fame). As a result, it helps you focus on decision-making rather than worry about the process of creating a chart or a plot, coding a macro or developing a simple predictive model. Why is this important? Let’s look at what many analysts have to do to get that chart or simple model.
Visualizations in R-complicated
I have always questioned why visualizing things in R and Python has to be so complicated. I can sense “R-ophiles” and hardcore coders moving on to a different blog with that statement. I’m going to ignore them and keep going. Shiny by RStudio is a giant leap in dynamic visualization capabilities; however, you need a significant amount of R programming to get a Shiny application up and running. For instance, here’s the code for creating a simple dashboard in Shiny.
If you’re not an R programmer or a data scientist, your eyes probably just glazed over at all that complicated and daunting code. Not only that, but all this code generates is a histogram with a slider.
Not to belabor my point, but even though Python has the matplotlib package, it is no walk in the park, either.
And, for all that hard work, this is what you get.
But we’ll always have spreadsheets, right?
Microsoft Excel is fairly simple to use and has already been a pioneer in descriptive analytics. But, creating macros and using complex pivot tables can be equally challenging for a typical business user. Some of these factors put analytics out of reach, which is driving up the demand for data scientists and analytics professionals. Clearly, what I have shown you here is not for the faint of heart. Do things really have to be this complicated to get a bar chart or a histogram? The answer is, absolutely and unequivocally, no!
The game changer for this data scientist: Dashboards
Sure, I agree that serious statistics, optimization, and programming skills are necessary for wringing deep insights out of data. However, plotting charts, producing graphs and creating some smart dashboards for business use should never be complicated. Otherwise, you defeat the purpose of analytics.
Watson Analytics is a game changer when it comes to creating dynamic dashboards. The tool is capable of deep exploratory analytics. Natural language processing features (a commonality this product shares with the IBM Watson ecosystem) enable quick drill downs.
Watson Analytics can also be used to create some basic predictive models and analyze Twitter sentiment. My opinion might be biased toward Watson Analytics. Nevertheless, we should all advocate for products, open source and otherwise, that can dramatically slash time needed for analytics.
Using Watson Analytics is as simple as click>load data>plot charts. Anyone can build dynamic dashboards with a few simple drag-and-drop actions. For example, this dashboard took about 5 minutes to build.
So, anyone can get new insights about correlations in data with this kind of visual presentation. Meanwhile, the data scientist can use the initial views and insights to formulate more complex models by means of tools like R, SPSS Modeler, and Python.
To drive the democratization point home, Watson Analytics is available as a free version with the only major limitation being the size of data you upload and analyze. The Plus and Professional subscriptions offer even more features (as you would expect; it can’t all be free right?). Consequently, gone are those days of mind-numbing slide presentations in endless meetings. With the dynamic dashboard capabilities in Watson Analytics, you can create quick drill-down charts and graphs on the fly. You and others can work collaboratively during team meetings, generating insights and making decisions.
Trying Watson Analytics
IBM Watson Analytics has ushered in a new era in democratizing analytics. I believe this is how IBM stands apart from other companies. You can learn more about Watson Analytics and sign up to try the Professional Edition and Watson Analytics for Social Media here.
About the blogger
Dr. Vishnu Nanduri, Ph.D. is a Data Scientist and Analytics Leader in Singapore. A former analytics practitioner at IBM, he also writes about data science and analytics. You can read more of his blogs on his website.