How to apply Insights in Excel and what to look out for when you do

Excel’s Insights feature uses artificial intelligence and machine learning to point out patterns in data that can aid decision-making, but there are several caveats to consider along the way.

Microsoft is gradually rolling out a new feature for Office 365 subscribers called Insights in Excel. This new tool has been in preview for Office Insiders for a few months but has now been released to everyone. The tool promises to apply the power of cloud-based AI and machine learning to Excel data sets to reveal new patterns and hidden information you may have overlooked.

Once Insights is installed, accessing it is as simple as clicking a specific button on the Office ribbon. However, getting the tool to actually provide useful information is a bit more problematic. Insights show both the promise of AI and its elusive effective practical application.

This brief tutorial article shows you how to implement Insights and explains some of the difficulties you may encounter when trying to apply AI to real-world data sets.

See: Artificial intelligence: trends, obstacles and potential gains (Tech Pro Research)

Excel Insights

To apply Insights, highlight your data set in Excel (give it a label if you prefer) and then go to the Insert tab on the Office ribbon. Click the Insights button, which is set by default in the middle of the ribbon (Figure A).

aexcelinsights.png

Figure A

When you click the Insights button, Excel will take a minute to analyze the data set and then present you with several graphical options (Figure B). The quality and quantity of these choices will of course depend on the data being analyzed. In our simple example, the options are fairly basic and not really all that enlightening.

bexcelinsights.png

Figure B

Assuming you find a graph that meets your needs, you can click the Insert Chart link and the graph will be inserted into your already formatted workbook in a presentable, albeit rudimentary form (Figure C).

cexcelinsights.png

Figure C

Warnings, warnings and warnings

It is important to realize that Insights works best with clean, well-organized data that is free of nested, empty rows and columns. Equally important, this tool cannot be used on data that has been converted to a pivot table. However, according to Microsoft, Insights will work best with data that is formatted as an Excel table (figure d).

dexcelinsights.png

figure d

To format a data set as an Excel table, highlight the data and use the keyboard shortcut CTRL-T. Although converting to table format, in our example, did not provide any additional options for the layout or hidden pattern detection, it might be more effective on a more complex table.

And that’s the problem with Insights and any other tool designed to use artificial intelligence or machine learning – it doesn’t always work. Proponents of these new technologies tend to overestimate the true capabilities of artificial intelligence and machine learning. Insights has tried several sets of real Excel data and the tool has not been able to even produce a single chart.

See: How to Use Excel’s What-If Tools to Analyze Business Scenarios (Free PDF) (TechRepublic)

Let’s take a look at our sample dataset. James is clearly associated with several huge spikes in activity and the charts reveal it, but do we really need Insights to see it? Of course, the tool cannot do anything to explain why the activity is so high. Someone will have to explore the reasons, artificial intelligence cannot help us in this regard.

It is important to keep in mind the overall limited capacity of what an AI can do in a working environment as it becomes more pervasive and the noise around it becomes louder. While AI may be able to provide a clue towards a pattern or solution, understanding the meaning and significance of that pattern will remain human responsibility for a long time to come.

Read also:

your thoughts:

Are you using artificial intelligence to analyze data? How effective is that? Share your thoughts and opinions with your colleagues at TechRepublic in the discussion topic below.

istock-664599098.jpg

Getty Images / istockphoto