BI without AI is like corn flakes without the milk
Latest update ... The link to the webinar recording and its corresponding Q&A of this article can be found here.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Certain pairings go so well together: cereal & milk, peanut butter & jelly, wine & cheese, soap & water, or spaghetti & meatballs. In fact, you couldn’t think of having one of these without the other. So too should hold true for your business intelligence, reporting, dashboarding and data visualization workflows (aka BI). Artificial intelligence techniques (or Augmented Analytics) democratize advanced analytics to more users by addressing business problems with automated quantitative insights (what I'm labeling as AI within this article).
The latest release of SAS Visual Analytics 8.5 for Viya 3.5 offers more than 10 capabilities that fall into the AI-enhanced business intelligence category. Some of these features are available in earlier releases of SAS Visual Analytics 8.x for Viya 3.x. A few are even available in even older releases (as noted below). Here’s a rundown of the features, in no particular order.
1. Data prep automated suggestions
AI can automate data preparation, which is the first step in the BI workflow process. Automated data preparation functionality may suggest table joins and other transformations, potentially based on data profiling or usage frequency. With a SAS Data Preparation license, you can use the Suggestions feature in SAS Data Studio. The Suggestions feature uses machine learning models to analyze your data and suggests transforms based on the type of data found in your data set as outlined by Mary Kathryn Queen. An example run-through of creating and running a data preparation plan that uses suggestions can be found here. For a more detailed tutorial, please watch this video.
2. Automated explanation
As Anna Brown, SAS Community Manager, laid out in her post about automated analysis, “You come into work and are told to analyze a completely new data set to find specific trends or insights that you, your team or your manager can use immediately to make a decision ... This scenario is where the automated analysis feature ... can help. Through machine learning, it tells stories about the data for you.” The output results are in a natural language-like construct as if a human had generated the visuals and written the findings. Read about the algorithm for Automated Explanation Objects in the documentation.
3. Automated prediction
Automated prediction takes automated explanation to the next level by allowing you to do point-n-click what-if analysis on the underlying factors for model building. The tool will build multiple machine learning models for your data based on your selected response variable. This is a fast and simple way to gain deeper insights. Watch a recorded demonstration by Jeppe Deigaard of how to take advantage of this capability, and where to find documentation.
4. Automatic charts
SAS Visual Analytics automatic chart display uses guided analysis functionality that automatically creates charts based on the data item(s) selected by the report builder . It’s more AI-enhanced if the software itself can recommend to the report builder which charts should be considered, based on analytical algorithms. SAS Visual Analytics does this as well – courtesy of the Suggestions Pane.
5. Related measures
SAS Visual Analytics also uses guided analysis to automatically depict potential relationships between measures based on data selections by the report builder or viewer through Correlation Matrices. It’s more AI-enhanced if the software itself can automatically and pro-actively depict potential relationships between measures in your data that’s part of the report. SAS Visual Analytics does this as well directly within the Data Pane itself, courtesy of Related Measures.
6. Text topics and sentiment analysis
A topic is a machine-generated category, the purpose of which is to indicate what documents are about.
Sentiment analysis is the use of natural language processing computational linguistics and text analytics to determine the attitude of a speaker or writer with respect to a topic document or other item of analysis.
Topics and sentiment analysis aren’t new features as they’ve been around since the v7.x for SAS 9.4 releases of SAS Visual Analytics. But v7.x release users who haven’t upgraded to 8.x yet, will be happy ... speaking of sentiment ☺☺☺😊... to know that these features – or any analytical visual object for that matter – can now be integrated (i.e., target of actions/interactions) with reporting-type objects, as depicted in this example. This is a much bigger deal for text topics and sentiment analysis, as only the word cloud portion was interactive within dashboards in v7.x.
Additional updated capabilities have been introduced within the 8.x releases such as:
- Additional languages support (now 30+), depending on license.
- View the full text of your documents in a pop-up window or in a list table object.
- Create derived topics data items from the topics list in your results.
7. Derived items
Derived Items are preset calculations (automatically provided by the software) that you can build off your existing measures. Note that this is also not a new set of features (it’s been around since the v7.x for SAS 9.4 releases of SAS Visual Analytics) ...
a) The list of automatic software calculations has been expanded throughout the history of SAS Visual Analytics releases.
b) Because the software is auto-generating simple to complex math or financial formulas with the click of a button and the user did not have to directly build the formulas themselves from scratch, I would like to promote this as AI-enhanced business intelligence. Even if it means to the point of debating it! ☺☺☺😉
c) People are always coming up with new or updated ways to use derived items to create custom report capabilities that are not automatically inherit within the software itself (albeit that is not the theme of this article). Here is but one example from Teri Patsilaras.
8. Demographic data
I personally call this the automatic “market research” feature. SAS Visual Analytics offers a plethora of geo spatial analysis capabilities (aka location analysis). A write-up of those would definitely be stretching the definition of AI-enhanced business intelligence … with one exception.
If you’ve created a radius-based selection (e.g., circle) on a Geo Map and you subscribe to and have enabled Esri premium services, then you can display demographic data for your selection. For example, you can display the total population and average income for the area that’s within a 10-minute drive from your place of business. ESRI Demographics support commercial market analytics and non-commercial usages, such as economic development, planning, and at-risk population assessment. ESRI maintains this data from a variety of providers.
Here are a couple of resources if you want to dig in here (for anything location analytics related within SAS Visual Analytics):
Get an introduction to ESRI integration with SAS Visual Analytics.
See more illustrations of adding geographic context to your categorical and quantitative data through location analytics.
Look at examples of what's new or improved in this area with the latest release.
SAS Visual Analytics provides a wide range of location analytics through native capabilities and integration with Esri.
9. Data-driven content or SAS Viya jobs
Data-driven content (DDC) is not AI-enhanced business intelligence for the report builder but in my opinion, it can be for the report viewers. As Renato Luppi said in his article on DDC, “At first, the name doesn’t tell much, but as you get to know more about what Data-Driven Content is and can do to help you achieve your reporting goals, I don’t think you will have another choice other than fall in love with it, as I did. So, what is it? Data-Driven Content (or DDC for short) is a ... report object ... that allows you to create your own JavaScript-based custom visualization and embed it in VA reports. Like other VA objects ... and others meant to be used with advanced analytics, DDC objects can participate as the source and/or target of actions (a.k.a. interactions).”
With these techniques, you will be able to extend SAS Visual Analytics with your own custom functionality to create the customized experience your team needs for business intelligence and data-driven decisions. The same holds true for SAS Viya Jobs. Please consult this white paper and video for more information. Although not everything one can incorporate with custom built web content classifies as true AI-enhanced business intelligence, there are examples that do – like this one about custom-built auto-generated related insights. I can think of many other unrelated examples, but that’s another article for another time.
10. Report summary
Report summaries are also not AI-enhanced business intelligence for the report builder but in my opinion, they are for the report viewers. As Melanie Carey explains in her article about dynamic report summaries, “Have you received ... a report and have no idea what the report is about or what information it contains? ... Wouldn’t it be more effective if along with ... the report you received an insightful summary? One that would quickly highlight any recent changes or key findings.”
SAS report summary helps to describe the report in a few sentences to replicate a speech-template. The functionality provides a dynamic description of the report: Conditional text, Dynamic values and Natural Language Generation ... a key requirement if you are going to claim this as AI-enhanced business intelligence. ☺☺☺ 😊 Further examples and details of how to utilize this feature can be found within this article written by Xavier Bizoux. Report summaries are also particularly useful if your audience includes individuals with visual impairments. The report summary can easily be read by screen readers. For a recorded demonstration of these and related screen reading capabilities, please watch this video.
Voice recognition on iOS devices
Speaking of voice recognition, not only can you easily access your SAS Visual Analytics reports on mobile devices (Apple iOS, Android, MS-Surface/Windows 10), but you can even use the speech recognition built-in to the device (Apple iOS) itself to issue commands, such as filtering the reports! Assuming natural language processing has been enabled on your system, there are various voice assistant commands you can speak into the device to interact with your report. For more information, please read the 'USE YOUR VOICE TO INTERACT WITH THE SAS VISUAL ANALYTICS APP' section within this white paper. For a recorded demonstration of voice assistant, please watch this video:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
What about for the true Data Scientist?
Automated machine learning can help even the most experienced practitioner. There are different levels of automation available now in the Model Studio environment of SAS® Visual Data Mining and Machine Learning (add-on to SAS® Visual Analytics). For full details, please consult this white paper and corresponding video from the SAS Advanced Analytics R&D team.
In addition, non-traditional approaches to forecasting can be represented using machine learning techniques. SAS® Visual Data Mining and
Machine Learning along with SAS® Data Preparation, SAS® Visual Analytics, and SAS® Model Manager can be instrumental in orchestration of a forecast generation process as summarized in this presentation by Mallika Dey from Core Compete.
Guidelines for determining when to use a machine learning model for forecasting are shown in this white paper and corresponding video by Steven C. Mills, where he depicts several neural network-based modeling strategies available in SAS® Visual Forecasting (another add-on to SAS® Visual Analytics) and the important factors to consider in choosing and training such a model technique.
You can even build your own Machine Learning web application using SAS AutoML. SAS AutoML offers the ability to automate the development of machine learning algorithms. This article will demonstrate the implementation of AutoML using banking loan applications as an example.
Building chat applications will even be possible courtesy of SAS Conversation Designer as depicted in this video.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Conclusion
Whether you perform AI (~ pour the milk) first or BI (~ pour the cereal) first depends on the data storytelling path you are trying to accomplish with your report community ... same with how much AI (~ milk) to amount of BI (~ cereal) ratio. ☺☺☺ 😊 Accommodate the user’s desired level of automation and assistance from the software. Expert data scientists may find AI enhancements to be a hindrance, while citizen data scientists may seek full automation. Regardless, any AI-enhanced business intelligence capability you can introduce to your user base, the more you can reduce time-to-insight and increase decision-making speed. Finding the perfect balance will determine how much your user community will adopt and value it (~ e.g. preventing too much left-over unused milk).
