Go To:
Artificial Intelligence in a Magic Quadrant
Microsoft, Qlik and Tableu have been listed among the leading providers of Business Intelligence systems by Gartner for years now. Solutions based on Artificial Intelligence are already offered by each of these leading companies.
- The Power BI tool, for example, uses the potential of Microsoft AI, enabling business users to prepare data easily, create machine learning models and draw conclusions.
- Tableau, which occupied second place in Gartner’s Magic Quadrant in 2020, uses Ask Data and Explain Data functionalities. They use natural language processing, Artificial Intelligence mechanisms and statistics to ensure better data analysis – both for business users and those with knowledge of the complexities of widely understood Data Science.
- Qlik rounds out the list in turn, introducing a functionality called Cognitive Engine in its April 2018 version, which uses Machine Learning mechanisms to suggest the most optimal visualizations, thus making it easier for users to analyze data.
SAP – Business Intelligence visionary
Gartner’s Magic Quadrant is a guide for business representatives when it comes to choosing a tool. It is worth observing the contest between so-called visionaries. Here Gartner indicates providers who know the market and have the potential to change it. SAP is one such visionary – and today we will take a closer look at what this well-known player has to offer. Since it is widely known that the cloud is a development direction for Business Intelligence tools, SAP has also decided to create something that will be a cloud alternative to the on-premises SAP BusinessObjects. The SAP Analytics Cloud tool (SAC), which has been available on the market since 2018, is currently the main direction of development for SAP in the field of data analysis and reporting. Let’s look at how SAP redefines work with analytical tools.
SAC functionalities
Data democracy has recently been a hot topic, promoting widespread access of users without specialized IT knowledge to unassisted data analysis, among others. Until now, such users were dependent on Business Intelligence specialists. Introducing the benefits of Artificial Intelligence to its analytical tool, SAP tries to meet the needs of users who do not have high-level analytical or IT qualifications. Let’s look at some interesting functionalities supporting data analysis that we can find in the SAP Analytics Cloud.
Search to Insight
Let’s imagine for a moment that we are analyzing sales data for beverages and we want to start with some basic information on sales results in recent years. According to the traditional approach, we would try to search for the report on our platform, hoping that it was titled in a way that is intuitive enough for us… But what about the possibility of using a search box in a different way, and – instead of looking for a specific report title – trying to ask a question which we want to find out the answer to?
Here are some examples of how Search to Insight uses Natural Language Processing to build reports based on search engine queries:
Query: “Gross margin time chart by location”
Answer:
Query: “Show me the top 5 products in Location Reno in 2016 Q1”
Answer:
The text query processing function works quite well, provided that we use the names of the dimensions available in the analytical model prepared beforehand by someone else in the queries. Fortunately, the user can display an available list of objects which they can use in their queries. Search to Insight works well for basic queries related to the aggregation of available metrics. However, if our queries exceed the capabilities of this tool, we can still count on AI support in the form of another module called Smart Discovery.
Machine Learning at SAC
The tool for report building in SAC, based on the existing data model, gives us the opportunity to use Machine Learning mechanisms. Machine Learning facilitates the discovery of patterns and statistically important relationships for our data source. When running such an analysis for a previously used sales model, the user only needs to indicate what element of the model will be the subject of the study (whether it will be a dimension or a measure), and then indicate which other elements are to be assessed in terms of the impact on the analyzed variable as part of the model.
For example, by indicating the gross margin measure as the main object of our analysis, we can get the following result:
Summary overview:
- summary presenting the trend of values in past periods together with forecast values for future months, calculated based on historical data
- distribution of values of a given measure presenting the most “typical” values for our measure
- analysis of the aggregates with the use of the dimensions available in the model
Key influencers:
- Indication of model objects that statistically have the greatest impact on the examined variable. In the analyzed model, the “Product” dimension was indicated as the factor with the biggest impact on the analyzed variable. In other words, the margin on selected products is very diverse.
Unexpected values:
- Identification of transactions in which the value of the examined variable differs significantly from the value forecasted by using the model. These types of values may be subject to further analysis in order to clarify or verify the correctness of transactions carried out.
Simple mechanism for “what-if” analysis. It allows you to assess the impact of changing particular model parameters on the variable you are studying, such as changing the discount value for selected products. It is a highly desirable functionality for business users, although in my view it still requires some refinement so that the presented results are more understandable to users.
The presented Smart Discovery tool is not the last functionality that SAC has to offer in the field of Augmented Analytics. After verifying the correctness of historical data, SAC can help us look to the future thanks to its Smart Predict functionality.
SAC predictive models
As part of predictive services, SAC offers three types of predictive models:
- Classification – used when the examined variable can have binary values. For example, we can try to predict if a customer will make a purchase (true / false) based on the available customer features.
- Regression – this model may apply a numerical value based on the diagnosed correlation between descriptive variables and the examined variable. This type of model can help us identify the descriptive variables that have the biggest impact on the result variable we are examining, and thus assess the potential numerical value of the examined variable for a hypothetical combination of descriptive variables.
- Time series – simply a time series forecast based on historical data.
The operating principle is similar for each type of model.
User:
- Preparation of historical data (file or database connection).
- Data import to the platform.
User:
- Indication of the column with the examined variable.
- Defining a filter for explanatory variables.
SAC:
- Division of data provided into a training set and verification set.
- Generating several predictive models and choosing the best by comparing the results with the verification set.
User:
- Division of data provided into a training set and verification set.
- Generating several predictive models and choosing the best by comparing the results with the verification set.
When assessing SAC functionalities in the field of data prediction, we can claim that their main advantage is simplicity of use, which will allow you to undertake complex analysis with no need for advanced statistical knowledge. This is why this tool should not be compared with other Machine Learning tools available on the market, which have much more developed capabilities, but therefore require a more in-depth knowledge of statistics.
Are you in the process of Digital Transformation? Get to know our services!
Do I still need Business Intelligence specialists?
After finding out more about the capabilities of Business Intelligence tools which use Machine Learning and the possibilities they open up for users, the following question may arise: “Do we still need Business Intelligence specialists?” My answer is: definitely! The competences of Business Intelligence specialists will still be necessary to create analytical models in such a way that will allow for their proper use in the above-mentioned Self-Service Reporting tools in modern SAP Analytics Cloud systems. Specialist BI knowledge will also be priceless in terms of integrating data from many different sources and building more complex analytical applications.
Summary
Self-Service Reporting tools equipped with the power of Artificial Intelligence uncover new possibilities for users who are looking for Business Intelligence tools. And while we’re still not using the full potential of Machine Learning, one thing is certain: AI is the future of BI! Business Intelligence specialists equipped with new tools will be able to provide consistently better business solutions. If your organization is facing new analytical challenges, take advantage of the extensive experience of Business Intelligence specialists to help make full use of the potential of your tools and achieve your business goals.
On daily basis, he supports clients in designing and building analytical solutions. In his career so far he has implemented projects using such platforms as Tableau, SAP Analytics Cloud, IBM Cognos, MS SQL Server. In private life, a fan (practicing) of running and basket.