Data Presentation in Business Intelligence

Introduction

In last three articles (1, 2, and 3) of this series, we discussed the overall Business Intelligence system architecture, different components, and how ETL helps store source data in the required form, followed by how OLAP (CUBE) helps to process, aggregate, and store huge amount of data. In this article, we’ll discuss another part of the overall BI system architecture: information delivery. How can we visualize or deliver data to help business owners and other users in the community? You can help them to know business trends and usability of the system, including other benefits. In the next the part of this series, you’ll know more about information delivery and various ways to represent data to help business owners in decision making.

Importance of Data Presentation

With the basic understanding of Business Intelligence, we should know the importance of data presentation in information delivery and how useful it is for business users. After storing data in the required form, either in a data warehouse/data mart database or an OLAP cube, data can be showcased in certain forms that will become the source of required information for all BI users; this is known as Reports in the BI world. These Reports can be data reports, executive dashboards, or conference presentations.

Data Reports may have different types of layouts to present the data, such as Metrics, Chart, Funnel reports, animated representation of data, and so on. All these kinds of data layouts can be considered, based on the targeted audience.

All these reports may represent Key Performance Indicators (KPIs) to visualize/present the data. These KPIs are a type of performance measurement. An organization may use KPIs to evaluate its success or progress, or to evaluate the success/progress of a particular business in which it is engaged. This data presentation is the logical end point for any BI system and becomes the source of truth for decision makers.

Reporting Tools

A reporting tool is an interactive data exploration and visual presentation of data that is stored in a database or cube and can be achieved by using various available reporting tools; the selection of a reporting tool for your purpose is solely your discretion but with supporting facts like we discussed, some selection criteria for OLAP tools in the previous article. There is no thumb rule to adopt a reporting tool; I want to reiterate this fact that the identification and adoption of any BI tool is not a personal decision and can’t be finalized without thorough study. You need to refer to the previous articles of this series to know more about the reasons why this is important for the successful implementation of any BI system.

There are varieties of reporting tools available that can be useful to present data and deliver the information. Some of them that come with Business Intelligence suites are Microsoft SQL Server Reporting Services, Oracle Reports, IBM Congnos Report Studio, and MicroStrategy Reporting Suite. Some open source tools are BIRT, JasperReport, Pentaho, and few other available tools for reporting are Tableau, QlikView, and the like.

Reporting tools help to visualize data in multiple, interconnected ways by using data regions. You can display data organized in tables, matrices, or cross-tabs, expand/collapse groups, charts, gauges, indicators or KPIs, and maps, with the ability to nest charts in tables.

Types of Reports

Once we finalize the reporting tool, the next step is to select the report design and data display to present the data as information for the user community. We need to consider key factors to design any report; these factors are the structure and tone of the report, length of report, kinds of data to include (tables, figures, graphs, maps, or pictures), detail of information, data positioning in the report, and the most important: visual sophistication.

We can explore different types of reports while considering the preceding key factors and organizing data in a variety of ways to show the relationship of the general information to the detailed. We can put all the data in the report, but set it to be hidden until a user clicks to reveal details; this is a drilldown action. Data can be displayed in a data region, such as a table or chart, which is nested inside another data region, such as a table or matrix. Another way is display the data in a subreport that is completely contained within a main report. Or, you can put the detail data in drillthrough reports, separate reports that are displayed when a user clicks a link. We’ll explore those ideas next.

Drilldown Reports: A Drilldown report provides plus and minus icons on a text box; these icons enable users to hide and display items interactively. This is called a drilldown action. For a table or matrix, it shows or hides static rows and columns, or rows and columns that are associated with groups.

Drillthrough Reports: A Drillthrough report is a report that a user opens by clicking a link within another report. Drillthrough reports commonly contain details about an item that are contained in an original summary report.

Subreport: A Subreport is a report item that displays another report inside the body of a main report. Conceptually, a subreport in a report is similar to a frame in a Web page. It is used to embed a report within another report.

Nested Reports: A Nest report can nest one data region, such as a chart, inside another data region, such as a matrix, typically to display data summaries in a concise manner or to provide a visual display as well as a table or matrix display.

A report can apply to a specific type of report item, a layout design, or a solution design. A single report can have characteristics from more than one type; for example, a report can be, at the same time, a stand-alone report, a subreport referenced by a main report, the target of a drillthrough report in a different main report, and a drilldown report.

Data Sources for Reports

Reports can leverage stored data in databases in the form of DW or DM or a cube to display. Data can be exposed by using a data source and data sets property in any report. These data sources and data sets can be exposed only to one report or shared with multiple reports. Report definition is always independent with the data source and data set and can be managed separately.

Different types of reports allow you to anticipate different report properties, such as drillthrough actions, expand/collapse toggles, sort buttons, Tooltips, and the use of report parameters to enable report reader interactions. To control data display or user interaction, we can use report parameters that can be combined with expressions and provide the ability to customize how report data is filtered, grouped, and sorted.

Industry Trends in BI Reporting

Industry trends are changing rapidly and user expectations are also growing with respect to information delivery by a BI system. In the current data world, a user not only analyzes structured data but also wants quick information on top of huge, unstructured data. We need to adopt and support new trends to meet user objectives in a rapidly changing paradigm.

We are observing great change in data visualization; the initial visual data discovery releases from the big organizations like Microsoft, SAP, IBM, Microstrategy, SAS, and Oracle tended to have limited capabilities, but the gap is slowly closing. The specialty organizations and the heavyweights are trying to find the right balance between analysis and trusted data.

Another important trend is mobility. Initially, the mobility of information was “nice to have,” but now it is an absolutely essential component of any useful report. The first mobile BI products allowed users to look at data remotely on their tablets and mobile phones, and as devices improved, dashboards and other visual representations made it easier for people to access the information they needed in the format they desired.

Another new trend is to present enterprise-wide massive amounts of data (Big Data). Everyone is talking about big data, but in the business intelligence world it means one thing: Organizations are being overwhelmed by massive amounts of new information that they need to analyze quickly and accurately. In the current industry trend, unstructured data will be a big part of this change because the ability to look at information not stored in spreadsheets and databases is letting organizations analyze information that they couldn’t truly leverage a few years ago.

It’s always highly recommended that organizations should change their way of operating and align with industry trends for better results and use the best of technology in their own interest.

Summary

In this article, we discussed how data visualization plays an important role for the successful implementation of a BI system. We discussed various types of reports and the factors that help in decision making for designing and implementing any report. Overall Information delivery is the last layer for a BI system, but this is the data presentation layer that turns data in information. It helps a diverse user community leverage to present the information; otherwise, users won’t be in a good situation to make any decision.

Visualization techniques revolutionize modern business intelligence gathering

Data scientists and analysts love digging into the architecture of data to grasp its essence, exploring how it works and divining what secrets it may hold. For some expert users, the very complexity of the data is what provides the “thrill of the chase.” However, the average user wants data that’s easy to understand. Visualization has proven to be the best way to make this happen.

Why does visualization work so well?

Noah Iliinsky, data consultant and coauthor of Beautiful Visualization and Designing Data Visualizations, spoke at a LinkedIn Tech Talk about the reason visualization is so powerful for analytics. “It turns out that our eyes and our brains have very sophisticated software built into them for things like pattern recognition and detecting when there are pattern violations on a variety of factors in terms of position, skew, color, size, blur, shape, etc. They are called ‘pre-attentive properties’. We can detect very quickly when something is different or out of position. If you leverage these well, you can design things where you can get a lot of information into someone’s brain very easily and very quickly.”

Enterprises are demanding more visualization

Business users may not know the science behind the way their brains process data, but they know what works. Presentation is king. Distilling data into the essential intelligence that will inform business decisions is pointless if the resulting reports are visually opaque. According to a recent TDWI white paper on self-service BI, “For information consumers, the results need to be easier to consume and use, and the solution here is to employ more sophisticated visualization techniques. These vary considerably, from using technologies such as Google Maps to display location-specific data, to visualization approaches such as small multiples, scatter plots, heat maps, enclosure diagrams, node links, arc diagrams, and more. Advanced visualization ranked third highest in the survey for enhanced user interface requirements, with 41% of respondents saying this was a ‘very important’ requirement.”

Tools must match the source, complexity, and variety of data

Being able to look at the same data in many different ways is critical since each perspective can add depth. The data visualization tools of the past, with their two-dimensional pie and bar charts, simply aren’t refined enough to offer real insight into complex data sets. Imagine trying to track the proliferation of power stations across the United States over time using a traditional Excel spreadsheet and a set of static graphs or charts. Assuming you have the relevant information stored on your SQL server, you could sort and present the data by date of initial operation, by state, by county, by power station type, and so on.

It turns out that our eyes and our brains have very sophisticated software built into them for things like pattern recognition.
Noah Iliinsky,
coauthor of Beautiful Visualization
However, making sense of the data really requires a map—and some way to visually express the changes that are taking place over time. The new “GeoFlow” visualization tool from Microsoft is a good example of how geographic data can be viewed in a way that permits the eye to easily detect patterns. It also includes the ability to drill down into the data after the overall trends become apparent so that users can uncover additional intelligence.

More features of a smart enterprise BI tool

Beyond offering many ways to present BI, the right solution will also give users more control over reporting. While visuals are important, they shouldn’t distract from the data or from business objectives. Every feature should be easy to use and fill a functional role. Here are a few key features that make a difference:

1. Interactivity, especially the ability to slice and dice the data
2. Full OLAP support
3. Static and dynamic capabilities
4. Visuals that relate to the real world within which the business operates
5. Multidimensional analysis
6. Collaboration for team analytics
7. Real time or near real time capability for live BI needs
8. Meta-visualization in dashboards

Visualization is about more than individual reports designed for distinct purposes or for certain departments. Dr. Joseph Morabito, Industry Associate Professor at the Stevens Institute, maintained in his 2012 talk about Big Data that different users require different types of visualization in terms of dashboards.

“The strategic dashboard is focused on high-level measures of performance. Typically, they feature static snapshots of data on a daily, weekly, or monthly basis, and there is little user interaction. You don’t want too much here. It’s better to be simple. Analytical displays are designed for detailed data analysis. Here, you’re going to have comparatively more data (and more complex data) but richer comparisons. You’ll have extensive historical data, but still mostly periodic snapshots. You’ll have a lot of interaction here with many OLAP features. Operational data requires a dynamic environment where we are using real-time or near real-time data (as in monitoring a supply chain management system). Here we need to keep it simple, as we do with the executive dashboard, but for different reasons. We need to see problems right away and then drill on demand so we can locate problems as they arise in real time.”

In the final analysis, users want more than BI solutions that enable them to achieve goals in their business. They want tools that make them feel smart. That’s the kind of positive reinforcement that provides the motivation leading to innovation. A data visualization solution that is versatile, well-rounded, and accessible for self-service is the best BI tool for this purpose.

Data visualization helps to bring data ‘alive’

Picture this: an easy-to-digest, up-to-the-minute window on how your business is performing. To some, it may sound like the panacea for the over-burdened executive desperate for details on their key targets. For others it may merely conjure up images of dispiriting data dashboards; technology that has promised so much yet delivered so little.

Now, however, a new breed of advanced data visualisation tools are arriving that claim to make good on those pledges of instant insight.

Part of the problem for many of the firms that wanted ready-made business performance information has been an over reliance on the wrong tools, says Jeremy Pile, technical director at food supply chain firm Muddy Boots. “By the time you’ve built a report using traditional business intelligence tools, it’s out of date,” he says.
Muddy Boots acts as an intermediary between food suppliers and the UK’s leading retailers, including Marks & Spencer. For example, if a retailer wants to ensure the quality of the food from its suppliers, Muddy Boots provides the data from the field to the shelf that shows how the produce has been cared for.

To help bring that data to life, Muddy Boots relies on data visualisation tools from software maker Logi Analytics. “Customers want us to present information that shows them how any given supplier is performing at a given time. They may want to know if that performance is consistent with previous performance. Having visualisation tools makes that much easier,” says Pile.

One aspect that sets the new breed of data visualisation tools apart from their BI precursors is that they provide different ways to present data, says Eddie Short, EMEA head of data and analytics at consulting firm KPMG. “Information is beautiful,” he says. “There have been analytical tools for as long as we’ve had computers, producing charts, and so on. But what we’re seeing now are tools that make it simple to understand the data.

“In the past, business reports were written by financial analysts and consumed by them and their management,” he adds.

Such reports might typically rely on bar charts, line graphs and the like; firms using advanced data-visualisation tools are more likely to be using bubble charts, geospatial heat maps or even word clouds.

Word clouds

Business intelligence purists might be sniffy about word clouds, but they turn out to be an excellent way of rapidly conveying important data, says Muddy Boots’ Pile. “For example, if one of our customers suddenly starts seeing a word like ‘packing’ appearing prominently in their word cloud, it might give them the first inkling that there could be a problem,” he says.

But there’s more to advanced data visualisation than mere fancy new graphs, says Stephen Few, principal analyst at Perceptual Edge. Users are beginning to expect to be able to ‘play’ with charts, recompile them on the fly or drill into the data.

That in turn requires a set of tools that can pull data from disparate sources – not just structured sources but, potentially, a myriad of unstructured sources. And, yes, the tools may frequently even be required to bring sense to that over-hyped buzzword, big data.

Data visualisation at Volvo

At Volvo, Andy Johnson, a support and project analyst at the car maker, has been using BI tools to monitor various aspects of business performance for many years. These were systems initially based on green screen systems, he notes. Over the past 15 months, Volvo UK has begun to explore ways of using data visualisation tools to monitor performance using much larger data sets.

The firm has always received sales data from the UK’s automotive trade body the Society of Motor Manufacturers and Traders, which helps Volvo track its performance against competitors. But historically, much of that data has been superfluous. “We target a fairly small fraction of the overall automotive market,” says Johnson.

Such practices are a common feature of trying to get business value from massive data sets, says Perceptual Edge’s Few. “At any given moment, most of the data that we collect is noise. This will always be true, because signals in data are the exception, not the rule.” But it is the ‘signals’ that give business leaders the insights they crave.

The analytic and data visualisation tools, QlikView, enable Volvo not just to zoom in on the market segment it’s competing in but compare how it is performing across the country, says Johnson. “Now, if we notice that sales have fallen unexpectedly in a particular location we might drill down and notice that one of our rivals has been offering a promotion in that region,” he explains.

But while the ability to create fancy charts using data from multiple sources is characteristic of advanced data visualisation tools, these are still capabilities that many traditional BI suppliers will claim for their products, says Short. To understand what really sets the new tools apart, you need to understand their underpinning philosophies, he adds.

Whereas typical business intelligence tools were built from a financial reporting perspective, with a bean-counter-like fixation on data accuracy, today’s data visualisation tools have been influenced by the philosophies of human-computer interaction, says Short.

Human-computer interaction

Human-computer interaction (HCI) emerged in the early 1980s, as computer scientists began to infuse software design with ideas from cognitive science and human factors engineering. But while concepts of usability percolated into office productivity, it has made minimal impact on enterprise applications – at least until recently.

The impact of HCI on analytic tools may be most easily understood as yet another example of the “Apple effect” on the enterprise. As Volvo’s Johnson explains, senior executives don’t just want to be able to access data visualisations on their desktop, they expect it on their smartphone or on their tablet.

And here, users are also coming to expect business charts to embrace the same type of interactivity they’ve grown accustomed to with Apple’s pinch-to-zoom capabilities, which set the iPhone’s image-processing capabilities apart.

While traditional BI vendors might see this as a systems integration problem, HCI professionals approach it from a usability perspective. A prime example is the work presented by Mikkel R Jakobsen and his colleague Kasper Hornbæk, researchers in human-centred computing at the University of Copenhagen at the VIS 2013 conference held in Atlanta, Georgia in October 2013.

Their paper explored how well users could interact with data visualisations when using devices with displays ranging from that of a typical smartphone to mega-size multiple monitor type displays. It’s often thought that the larger the display, the easier it is for people to work. But what Jakobsen and Hornbæk showed is that the time people take to complete tasks – in this case using maps – can actually be longer on a large display.

Other speakers at VIS 2013 also challenged the perceived wisdom over data visualisation, including Michelle Borkin from the School of Engineering & Applied Sciences at Harvard University. Her research demonstrated that the more data visualisations included annotations and decorations, often dismissed as “chart junk”, the more memorable they became.

Such observations are testament to the difficulty of successfully creating data visualisations. “If a business wants to get real insight from data – and not simply drown in it – it needs someone who understands both what the data presents and how to display the results,” says Short.

In many ways, advanced data visualisation tools fall neatly in line with one of the big trends in BI over recent years: the democratisation of analytics; taking data tools outside the purview of financial analysts and bringing them to the masses.

But to deploy advanced data visualisation tools propitiously, however, firms need to be wary of difference between this approach to data analysis and more traditional business intelligence. Creating successful data visualisations needs some capable of both “left brain” and “right brain”-type thinking, says Short. It’s the ability to understand what the data means and how that can be presented – on whatever device – in such a way that it will immediately convey meaning to whoever looks at it.

“When we first started using our visualisation tools, we were quite prescriptive about how they should be used,” says Volvo’s Johnson. “Then we began to see the ideas other people had and it opened up our thinking about what could be done.”

But in can also be prudent to place a limit on users’ freedoms as not every method of visualising data is useful, says Johnson. “It can be easy to get carried away,” he adds.