Q&A in Power BI

With soci­ety demand­ing more nat­ural inter­ac­tions with com­puters and arti­fi­cial intel­li­gence, nat­ural lan­guage pro­cessing (NLP) has quickly become one of the most import­ant BI trends for 2018. With Apple’s Siri, Amazon’s Alexa, and Microsoft’s Cort­ana lead­ing the advance­ment of NLPs, we are becom­ing increas­ingly exposed to the oppor­tun­it­ies NLP sys­tems provide.

Unsur­pris­ingly, the busi­ness intel­li­gence envir­on­ment is the per­fect place to imple­ment NLP tech­no­lo­gies. The enter­prises of today need to not only glean insights into their busi­ness, they need to be able to inter­act with their data without hav­ing to depend on an SQL or visu­al­iz­a­tion tool expert to do the leg­work for them.

“By 2020, 50% of ana­lyt­ical quer­ies will be gen­er­ated via search, nat­ural-lan­guage pro­cessing or voice, or will be auto­mat­ic­ally generated.”

Source: Magic Quad­rant for Ana­lyt­ics and Busi­ness Intel­li­gence Plat­forms (2018). Gart­ner, Inc.

In this art­icle, the nat­ural lan­guage solu­tion offered by Power BI will be tested. Ques­tions and answers (Q&A) is an integ­rated fea­ture in Power BI that empowers users to be able to ask ques­tions and receive visu­al­iz­a­tions as a response.

It’s worth­while first tak­ing a look at how Q&A from Power BI works. Power BI reads the names of columns within the data­set cur­rently in use; there­fore, it is extremely import­ant that you struc­ture the dataset’s column names with func­tional mean­ings that are eas­ily under­stand by users. In the end, users will be con­nect­ing dir­ectly to the data­set pub­lished, so take cau­tion to avoid messy nam­ing con­ven­tions that are avail­able in some datasets.

For example, sup­pose you had an Excel table named “Sales”, with columns titled “Product”, “Month”, “Units Sold”, “Gross Sales”, and “Profit”. You could ask ques­tions about any of those entit­ies. You could ask “show sales, “total profit by month”, “sort products by units sold”, and more.

This Q&A fea­ture is also avail­able when you accede a pub­lished dash­board. In that case, you can con­fid­ently ask ques­tions about the meas­ures and dimen­sions dis­played. For instance, in the dash­board below, it is pos­sible to ask ques­tions about the ‘Gross Mar­gin %’ and the ‘Total Revenue’.

Also, when you click in the ‘Ask a ques­tion’ box, some options for the data avail­able appears. This can help dur­ing situ­ations where you are not famil­iar with the data­set that is provid­ing the data.

There is also a really help­ful auto­com­plete func­tion when ask­ing a ques­tion. This auto­com­plete sug­gests meas­ure­ment names, dimen­sion names and dimen­sion members.

Fur­ther­more, Power BI auto­gen­er­ates visu­al­iz­a­tions to dis­play the data. How­ever, you can choose which visu­al­iz­a­tion type answers your ques­tion in the best way. Power BI also gives you some sug­ges­tions when ask­ing, but bear in mind that, depend­ing on the data, not all visu­al­iz­a­tions are always available.

Res­ults

The fol­low­ing videos shows how to use the Power BI Q&A functionality.

Con­clu­sion

NLP is gain­ing pop­ular­ity in the busi­ness intel­li­gence world, ensur­ing that more products and func­tion­al­it­ies will appear cov­er­ing this need in the near future. Power BI ‘Q&A’ is a use­ful fea­ture that acts as one such solu­tion. Other busi­ness intel­li­gence tech­no­lo­gies like Tableau are also invest­ing in this sub­ject, most not­ably with the recent announce­ment of their acquis­i­tion of Clear­Graph, a com­pany that enables smart data dis­cov­ery and data ana­lysis through NLP. Expect more BI tech­no­lo­gies includ­ing this func­tion­al­ity in the fol­low­ing months.