In dit artikel nemen we de meest recente versies van de BI-pakketten Tablue (Tableau Professional 9.0), Qlik (Qlik Sense 2.0) en Microsoft (Power BI) onder de loep. Wat bieden de pakketten aan functionaliteit en hoe goed presteren de in de praktijk?
Computerworld selecteert hier de beste Engelstalige verhalen uit ons internationale IDG-netwerk.
Watching the new, easy-to-use, exploratory business intelligence products edge out the older, feature-bloated reporting platforms for systems of record is like watching evolution in action. Let's face it: The first-generation BI products required their own IT priesthood to set up, and they incurred the same implementation delays as every other IT project. Things got worse, not better, as the various BI product marketers got into feature wars instead of actually trying to deliver business value.
The three data discovery products I've been reviewing over the last few months -- Tableau, Qlik Sense, and Power BI -- demonstrate a range of capabilities at a range of prices, with Tableau the most capable and expensive, Qlik Sense in the middle, and Power BI the least capable, but a very good value. Each product might be useful to you, depending on your specific requirements and budgetary limitations.
Note that there are about two dozen significant products in the BI category; the three I discuss here are not your only choices. As you'll see, however, Tableau, Qlik Sense, and Power BI all offer a good selection of chart types and data sources, and they are all much easier to learn and use than the traditional BI products. On the other hand, other products are better for routine line-of-business reporting, even if they are not as good as these products for data discovery.
Tableau Professional can connect to a wide assortment of file (Figure 2) and server data sources, including Excel workbooks, character- and tab-delimited files, statistical files, and upward of 40 server types, although 19 of those are available only from Windows. Tableau Personal is restricted to six kinds of data source; the free Tableau Public can use a mere four kinds of data source.
You can connect multiple data sources to a Tableau worksheet and create joins between tables or files. If you know the joined data has referential integrity, you can improve performance by telling Tableau to assume referential integrity.
Tableau does a good job of importing data whether or not it's in an ideal format. It does a similarly good job of inferring geographical information from street addresses. Tableau also gives the analyst a lot of flexibility to pivot, split, reformat, and massage the imported data without having to return to the original data source.
Figure 1: Tableau 9's Welcome screen offers easy access to data connections, workbooks, training, and resources.
Tableau analysis is a drag-and-drop process with property sheets, kind of like a Visual Basic for data scientists. As we can see in Figure 3, the data dimensions (fields used for classification) and measures (data values both primary and calculated) appear in the tab at the left. You can drag them into rows and columns, attributes and filters.
Every feature of Tableau has additional options, even though the default settings are often pretty good. For example, the size of bubbles can be controlled in the Edit Sizes dialog that comes up when you double-click the bubble card, below the marks and colors cards to the left of the chart in Figure 3. Color, size, and shape give you the ability to represent extra dimensions and measures on a given chart in addition to the row and column measures. You can also do a lot with actions and tool tips.
Tableau lets you organize your analyses into worksheets, dashboards, and stories. Dashboards can contain an arrangement of multiple worksheets, and you can create actions for one worksheet that affect other worksheets in the dashboard, making it easy to, for example, see the data for a particular city over time, as in the "See what's happening in your home town" dashboard of the CNBC Recovery Watch story.
Figure 2: Tableau 9 can read a wide range of files (shown above) and servers. New in this version is support for SAS, SPSS, and R data files; Apache Spark SQL servers; and regular expressions in calculated fields for PostgreSQL, TDE, Apache Hive, and Oracle.
CNBC Recovery Watch is an example of a Tableau story on Tableau Public, which as you've seen displays the analyses in a browser. Anyone can sign up for Tableau Public for free, then download the Tableau Public app for Windows or Mac to create other analyses. That version of the app can only save analyses to the Tableau Public server.
Tableau Server is a private version of the system you recently used. It runs on Windows, has an administration system to control sharing, and can work with all the same data sources as Tableau Desktop. Tableau Online is a hosted version of Tableau Server.
If you want to share analyses privately but don't have access to Tableau Server or Tableau Online, you can export a "packaged" workbook from Tableau Desktop, send it to your colleagues, and have them open it with the free Tableau Reader.
Figure 3: Setting up a Tableau worksheet is even easier than creating a graphic or pivot table in Excel, and offers many more options both in terms of graphics and analytics.
Tableau's selection of chart types is very good, and Tableau provides excellent control over chart appearance. It also gives the analyst easy ways to display multiple dimensions and measures. Tableau's maps, dashboards, and stories help the analyst to explain the logic leading to a conclusion, and its parameterized displays controlled by widgets allow the viewer to play along.
Tableau supplies sample data, videos, quick starts, live classes, and webinars to help people get up to speed on the product. All versions of Tableau are available as free trials.
It's very easy to learn the superficial stuff in Tableau, and many people can become productive with it in half an hour. However, it's harder to learn features that are beneath the surface unless you happen to have seen them in a training video or know their names, so you can search the help.
Qlik Sense 2.0
When you first open or create a Qlik Sense "app" (essentially a project), you see an overview screen (Figure 4) that displays its sheets, bookmarks, and stories, as well as a project summary. From here you can view any part of the project or create a new sheet. The compass drop-down at the top left allows you to get to the data manager, data load editor, data model viewer, and the hub. The list drop-down lets you add data, perform sheet operations, and get help.
A Qlik Sense sheet shows one or more visualizations. The visualizations are automatically associated for selection purposes.
Figure 4. A Qlik Sense App overview displays its sheets, bookmarks, and stories, as well as a project summary. From here you can view any part of the project or create a new sheet. The compass drop-down at the top left allows you to get to the data manager, data load editor, data model viewer, and the hub. The list drop-down lets you add data, perform sheet operations, and get help.
You can select a dimension, measure, or a group of items from any visualization in a sheet (Figure 5). The other visualizations on the sheet will be updated with the same selection. You can also make multiple selections.
For presentation purposes you can organize sheet snapshots into stories and annotate the pages to bring out your points (Figure 3). If you need to drill down in response to a question about a story page, you can easily right-click on the relevant sheet and go to source. Once you are done looking at the source sheet, you can return to the story.
Figure 5. Here we have selected the Produce product group for preview by clicking it in the top donut chart. Notice that this selection shows up in the top left and has changed every chart and number on the sheet. Clicking the green check box will change the preview to a full selection and allow us to continue our analysis.
Qlik Sense 2.0 introduces the Qlik DataMarket, a source of public and commercially available data in six categories: business, currency, demographics, society, weather, and the economy. In each category, you will get some data for free, and all of the data if you have a Data Token. For example, in the economic data, you get the World Bank selected development indicators for free, but you need to pay to get the Federal Reserve Bank data on the U.S. consumer price index and the U.S. employment numbers.
Qlik Sense can import data from delimited files, Excel files, ODBC, OLE DB, and Web files (JSON with an add-on, plus XML and HTML tables). You can combine and relate all of these freely, and you can view the resulting schema in a data model view. Qlik Sense can connect to essentially any database, via ODBC and OLE DB, including "big data" sources such as Cloudera, Hortonworks, Vertica, and Teradata.
Figure 6. For presentation purposes, you can organize sheet snapshots into stories and annotate the pages to bring out your points. If you need to drill down in response to a question about a story page, you can easily right-click on the relevant sheet and go to source. Once you are done looking at the source sheet, you can return to the story.
Qlik Sense normally keeps all of its data in memory, in compressed form. There are times, however, that you have too much data to fit in memory. In that case, Qlik Sense offers "direct discovery" mode, which combines in-memory data with in-database data on demand. In direct discovery mode, some fields are loaded into memory only as metadata/symbol tables that can be used in expressions. The actual data residing in the database will be queried as needed.
Visualization types in Qlik Sense include bar charts, combo charts (bar plus line), gauges, KPIs, line charts, maps, pie (and donut) charts, pivot tables, scatter plots, tables, and hierarchical tree maps. Scatter plots include bubble size and color, but not bubble shape. You can add filter pane widgets and text and image areas to any sheet.
When working with Qlik Sense, you can save a bookmark to the current selection state of the current sheet. You can combine bookmarks into stories and add text and other annotations to make the story self-explanatory (Figure 6). If you're using a story for a live presentation, you can drill down to the source for any visualization to answer a question, then return to the story when you've answered the question.
You can share your Qlik Sense analyses in multiple ways. Even if you're running the desktop version, you can print a sheet or save it as an image file and send that to colleagues. You can share your desktop apps to the Qlik Sense Cloud, and up to four colleagues can explore your app. The Qlik Sense Cloud doesn't currently support creating or editing apps, but that capability is planned for the future. Paid expansion of the Qlik Sense Cloud to more than five users per group is also planned, but the pricing has not been determined.
The most complete option for sharing is Qlik Sense Server, which allows any number of users to create, edit, and share apps. At $1,500 per token per year, however, Qlik Sense Server is not cheap, although it's less expensive than Tableau.
I found the online articles and videos for learning Qlik Sense to be very good. As it turned out, I needed to spend quite a bit of time learning Qlik Sense -- there was a lot in the product that wasn't obvious to me at first glance.
Microsoft Power BI
With the introduction of the new stand-alone Power BI, currently in preview, Microsoft hopes to compete with and perhaps leapfrog next-generation BI incumbents Tableau and Qlik. The new Power BI includes a Web interface to a service hosted on Azure and a Power BI Designer application for the Windows desktop, and it's much more modestly priced: A standard account is free, a Pro account is $9.99 per user per month, and the Power BI Designer is free.
Both the website and the Designer application are updated on a regular basis. The Power BI Designer is updated monthly; it's hard to tell when the site is updated.
You can import data into Excel, into the Power BI Designer, and into the Azure site. The data can come from Access, Active Directory, Azure SQL, DB2, Excel, Exchange, Facebook, GitHub, Google Analytics, HDFS, HDInsight, Marketo, Microsoft Dynamics CRM, Microsoft Dynamics Marketing, MySQL, OData, Oracle, Postgres, Power BI Designer files, Salesforce, SAP BusinessObjects BI Universe, SendGrid, SharePoint, SQL Server, SQL Server Analysis Services, Sybase, Teradata, Web tables, Visual Studio Online, and Zendesk.
Figure 7. Power BI reports can contain multiple visualizations. Visualizations can be interactive, and interactions can affect the appearance of other visualizations.
For some data sources, Power BI has predefined charts, dashboards, and reports. For example, the default Visual Studio Online dashboard and report provide at-a-glance views of Git, pull request, and version control activity across the projects you configure for your account. For other sources, Power BI expects to see certain markers for its data. For instance, it supports Excel Worksheet named tables, Excel Data Model tables, and Power View sheets. If you only have raw data in your Excel worksheet, you need to go back to it and create one or more named tables; it also helps if you make sure your data types are correct.
If you upload a worksheet from your computer, Power BI will use it as a static snapshot. If the worksheet resides on your OneDrive, Power BI can refresh its data on demand. If you want to import text or CSV data files, you need to start by importing them into Excel or Power BI Designer.
When you bring data into Power BI, that data becomes a data set. A data set can be used in multiple reports, and visualizations from the data set can display on multiple dashboards. Similarly, a dashboard (Figure 7) can display visualizations from many reports and many data sets.
Figure 8. Power BI dashboards can display multiple visualization tiles. Asking a question (near the top) can create a new visualization.
A Power BI report created on the website can use data from a single data set; a report created in Power BI Designer (Figure 8) can join multiple data sets. On the other hand, some data sets, such as Power View sheets and SQL views, may look and act like they have already joined multiple tables. A parent-child relationship in the data unlocks one of the coolest features of Power BI, the ability to drill down into segments of the data interactively.
Power BI Designer (Figure 9) allows you to shape and transform your data in ways that the Power BI service can't. Once you define the shaping steps in the Designer, however, the service can rerun them every time it connects to the data.
In addition to transforming data columns, the Designer can merge tables, drop columns, and create new measures with Data Analysis Expressions (DAX). DAX is very much in the spirit of Excel formulas, only using named columns, not cell ranges. For example:
Previous Quarter Sales = CALCULATE(SUM(Sales[SalesAmount]), PREVIOUSQUARTER(Calendar[DateKey])).
Figure 9. Power BI Designer makes heavy use of contextual menus. Here we are changing the type of a column of data from text to whole numbers. This particular step could also have been done from the Transform ribbon.
In general, you have two ways to create charts in Power BI. You can select variables from a data set, then select a chart type, or you can ask a question in Q&A (natural language query). You can select fields by checking their box, by dragging them into the visualization area, or by dragging them into the correct bucket (axis, size, and so on).
Power BI -- both the service and the Designer -- can currently create some 16 types of charts from your data, although not all of them can be pinned to a dashboard or recognized by Q&A. The visualizations you can pin to a dashboard are bar and column charts (clustered, stacked, and 100 percent stacked), cards (with or without KPIs or images), combo charts, funnel charts, gauge charts, line charts, maps, tree maps, pie charts (with a legend but without details), scatter and bubble charts, stand-alone images, tables or matrixes (with no images or KPIs), and single-number card tiles (made from the question box or a card with a single measure). If you use a card, make sure you have set an aggregation in the Field Well rather than choosing the Do Not Summarize option. (You can view examples of the supported Power BI visualizations here.)
The Power BI app for the iPad (Figure 10) can view dashboards and drill down into reports. It also lets you interact with tiles on your dashboards, create favorites, share dashboards, and annotate and share a snapshot of a tile.
Figure 10. Microsoft Power BI for the iPhone and iPad can view dashboards and drill down into reports. The image shows a sample dashboard for a director of marketing.
From the Power BI site, you can share and unshare dashboards with people in your organization -- that is, people who have the same email domain or occupy the same Power BI tenant. They will have to sign into the service, using either the site or the mobile viewer, to view and interact with the dashboards. A free account is sufficient for viewers.
Colleagues with whom you've shared a dashboard can see your dashboard and interact with your reports in Reading View. They can't create new reports or save changes to existing reports. They do see updates once you save your changes. Colleagues can't see or download the data set, nor use any of the data refresh operations.
Learning Power BI at a superficial level is a snap, and the introductory learning materials are easy to follow. However, in true Microsoft fashion, parts of the product are obscure, rely on other Microsoft products, and are documented in forbidding complexity.
Analyzing a single Excel table is easy to do and easy to learn, but setting up Power View sheets in Excel and relating them to each other is tricky to do and not easy to learn. Importing data from a remote server may require you to consult your IT department, not only for credentials, but also for the correct driver and settings, as the Microsoft documentation is likely to leave you scratching your head.
How to choose?
Power BI is a promising preview of a self-service business intelligence system. While Microsoft is starting to get the idea that not everyone wants to use Excel for analysis, Power BI currently relies on Excel or the Excel-like Power BI Designer application for data conditioning and joins between data sources. This is acceptable for the price, but not as convenient as Tableau or Qlik Sense.
Similarly, Power BI usually does a reasonable job of setting the axes on visualizations and setting the sizes of bubbles (in bubble charts), but doesn't let you refine the appearance of the visualization. Again, this is acceptable for the price, but not as convenient as Tableau or Qlik Sense.
I like the way that Qlik Sense can connect to essentially any ODBC or OLE DB data source, Excel files, and most delimited text files, although the text data import is not as convenient as Tableau's. The way Qlik Sense handles Web data sources is pretty good. There are a few common systems of record that Qlik can't currently reach, but the company is working on that.
I appreciate the automatic indexing of Qlik Sense's associative, in-memory data set, as well as its facility for handling data too big for memory. I like its way of automatically wiring the visualizations on a sheet, rather than making the analysis define the relationships.
I like Qlik's associative green-white-gray experience in which colors of displayed values indicate state (selected-selectable-not selectable). This scheme helps you to spot both related and unrelated data without having to dig. I also like Qlik's way of defining expressions, but not quite as much as I like Tableau's. Qlik Sense offers good control over the appearance of visualizations -- better than Microsoft Power BI, but not quite as good as Tableau.
I found Qlik's support materials to be very good, but even so Qlik Sense was not as easy as Tableau for me to learn to use; it lags Tableau in both categories.
Qlik Sense is cheaper than Tableau but much more expensive than Power BI. That is consistent with its capabilities being intermediate between Tableau and Power BI. Considering the power and ease you get for the price, Qlik Sense is a good value.
Tableau has an excellent assortment of data sources ranging from Excel, text, and statistical files to database servers, cloud data warehouses, various flavors of Hadoop, and systems of record (such as Salesforce). Importing data is easy, as are data cleaning, inline data transformations, and construction of data joins.
Tableau's selection of chart types is very good, and it provides excellent control over chart appearance. It also gives the analyst easy ways to display multiple dimensions and measures. Tableau's maps, dashboards, and stories help the analyst explain the logic leading to a conclusion, and parameterized displays controlled by widgets allow the viewer to play along.
Tableau makes deep statistics available without writing code, although you can do even better if you do write code, especially R code. Overall it's not hard to learn Tableau at a basic level. Learning the finer points will take some time and patience, however.
Price can also be an issue for Tableau customers. With Microsoft and Qlik offering free options, it can be difficult to justify buying Tableau analysis capabilities for a whole company, although some companies have done so. That leaves the company with a trade-off: Give Tableau to only the people who absolutely need it to stay within budget, or buy a slightly less capable and much less expensive product for everyone.
Finally, note that reporting is an area where some organizations will find Tableau, Qlik Sense, and Power BI to be lacking. These products do a great job of self-service reporting and analysis, but they won't always meet the needs of IT departments for production line-of-business reports. Some organizations may find they have to buy and maintain a second BI product for these needs, either because they deem the data preparation capabilities of these products inadequate for scheduled automatic reports, or because they want a product that can do more sophisticated statistical analyses on a routine basis, and alert management appropriately when goals or guidelines have not been met.