The Future Analytics: Part 3 – Apps and Visualization
This is part three in a four-part series. You can find the other blogs in the series here:
The Future Analytics: Part 1 – Overview
The Future Analytics: Part 2 – Big Data, Predictive and HANA
The Future Analytics: Part 4 – Suite and Actions
Watch this video to learn more about the Future Analytics: The Future of Analytics & Big Data (sapserviceshub.com)
Here’s a remarkable fact: the iPad first came out in April 2010, and as of June 2014 there have been over 200 million sold. That’s roughly equivalent to the entire population of Brazil. In 2013 alone, nearly 200 million tablets were sold, with the iPad taking 36% of the market share. This is a remarkable growth in a platform that didn’t exist just a few years ago. If we add in smart phones, the numbers are even more staggering: in Q2 of 2014, over 300 million smart phones were sold world-wide. It has changed corporate IT landscapes dramatically, and BYOD or “Bring Your Own Device” is increasingly common.
This has also brought about a revolution in how users perceive software. Tablet and smart phone users have become accustomed to Apps, which are self-contained programs designed to fulfill a single particular purpose, rather than single, feature-rich applications that provide lots of functionality. Whether users use their tablets only for personal use or also in a corporate business environment, the expectations users have of software has changed as a result.
This has implications for business applications, which historically have not worked that way. SAPGUI, let’s be honest, feels from a pre-internet age, and initial SAP web applications seemed to be modeled on SAPGUI more than trends in web development. But no longer can we expect users to be satisfied to have to use multiple tools and applications to perform their daily business tasks, and in interfaces that look nothing like what they have accustomed to in their daily interaction with technology. Why would we need to have multiple browser windows and applications open to execute a particular business process, when we could have all that in a single location?
With SAPUI5 and Fiori Apps built with it, this changes completely. Whereas in the past software tended to define the business process (“This is how that works in SAP, and you’ll need to train your workforce accordingly and adopt these best practices”), now the software can follow the business process. The SAPUI5 web application framework allows us to quickly and easily build applications that adapt to whether they are viewed and used on the desktop, tablet or smart phone, and tie back into the Suite back-end. We’re already building out a large selection of such applications that can be implemented on existing Suite systems, but the framework allows us to do much more than that.
While the Fiori apps allow us to approve purchase- and sales orders, manage expenses, etc., this is not the limit. We can bring in additional information to further improve such apps, by bringing in past actuals or any predictive analysis to provide the user with information to make appropriate and well-informed decisions. In a retail environment, we could conceive of a shop manager walking the isles in the store, checking what’s still left on the shelves and getting through an app the current in-stock levels, past week sales and any trends in this, as well as likely sales in the coming days through predictions based on past history and additional information like expected weather forecasts, local events, and even advertised promotions in other stores. We could provide recommendations on how many new products to order, and in case we receive electronic bids from suppliers could even rate different suppliers against each other to fulfill such an order at the desired price and volume, as well as taking things like distance from distribution centers and how quickly the order can be fulfilled into account. All in a single app.
As this example illustrates, we can now build apps that provide users with the tools to perform their daily tasks wherever that may be. Business software should support the core business activities and not get in the way or require a lot of training to be able to properly be used. Obvious, well-designed and easy-to-use applications (who needs to explain an app or reads the manual, ever, if they even have one?) that focus on a unique task while providing all the background context and information to make appropriate decisions will create further efficiencies in business operations while allowing users to focus on the things that truly matter.
Whether building an app that integrates with the Suite and includes analytics, or even dedicated analytics apps, we need to take a second look at how we visualize the data, though. What works in Excel or a Business Intelligence tool on the desktop, may not work well on a mobile platform. Wide tables with many columns don’t work well on tablets, let alone smartphones. Now that a lot of SAP’s senior leadership operates off of tablets, we are now implementing new analytic apps based on SAPUI5 to provide a snapshot insight into multiple aspects of the business, and the first reactions have been very positive.
You can see an example of such visualizations in the screenshot below (using sample data). These charts are not part of the SAPUI5 toolkit – although there is a visualization library, it is not yet complete and still in flux – but the framework allows for the creation of custom controls that you can subsequently use inside your own apps. You can find a more technical description around that here, but the key point is that you can design and create your own D3.js charts, giving you all the freedom to implement any visualization you think would best communicate what you’re trying to say, and visualize aspects of the data that otherwise might not be as simple to get from a table of numbers.
The scatterplot in this screenshot can serve as an example for this. The data was originally displayed as a table with revenue by segment across multiple quarters, and included the deal count. This scatterplot shows for each quarterly data point both total revenue for a segment, as well as the deal count. We show the average deal size as well (across the entire dataset underneath the legend, and in the tool tip for each data point). This gives us a way to show from what segments the revenue comes from, but also how the segments compare to each other in terms of average deal size.
Thinking through how your data needs to be visualized becomes even more important once we bring in Big Data and predictive analysis. The main added complexity is that we have to express uncertainty in a way not to mislead with an representation more definite than the analysis can justify. In particular, these would be confidence intervals in forecasts or around the slope of a regression line. Charts using those are common in statistical packages, but are rare and often insufficient in most BI tools and even Excel. However, such graphs can be rather easily designed in D3.js and used within the framework, as you can see below in a forecast with 80% and 95% confidence intervals.
In the final episode in this series, we’ll discuss Suite integration and recommended and autonomous actions. You may also be interested in this blog on visualizing uncertainty.
To learn more about how SAP HANA Services can help you throughout your Analytics journey, please visit us online.