You'll probably frown your eyebrowes when I say that SAP Analytics Cloud is truly changing the analytics landscape. "You work for SAP, Iver ... of course you say that" ..... well, that's correct but still my statement will hold. Apart from it's unique core roots - combining Analytics, Planning and Predictive in One tool -, it's embedded data point commenting, it's end to end collaboration, there are now the staggering Machine Learning capabilities.
The embedded Machine Learning capabilities in SAP Analytics Cloud are all clustered in so-called Smart Features. There aim is to guide and assist the user to get better and more valuable insights in a shorter time. 3 Keywords here:
Better: find insights that are not immediately obvious using correlation algorithms. For example key influencing attributes to a KPI
More Valuable: run automated simulations and find exceptional values automatically. Add this correlations to your dashboard and stroyboard instantly
Faster: apart from the algorithms finding insights that would normally take a lot of time, you can now "just ask" SAP Analytics Cloud for insights just like you do on Google. Oh, and if you really don't knwow here to start, the product can oan so auto-generate complete dashboards for you in split-seconds
The Smart Features are already there for quite a while, yet some of them are bettered significantly. In this blog I will highlight the bettered Search to Insight and the new Smart Discovery 2.0. Both are part of SAP Analytics Cloud.
SAP Analytics Cloud starts analyzing for you
We get to a stage where the software is analyzing for you instead of with you. And it does that in a remarkable good way. Imagine how valuable this is when you as an analyst require to understand a new data source. You connect, run the Smart Feature and instantly get a thorough understanding of the inights that really do matter. An imagine the value when you data source is updated massively and you need to under business impact immediately.
At Sapphire 2018 many important announcements and innovations were being made amongst numerous once for SAP Analytics and the Smart Features. All Sapphire 2018 presentations can be reviewed here.
Search to Insight
We all use Google, right? It has seach an intelligent way of searching. I can just ask "how many people live in belgium in 2017" and I get instant results carefully ranked by best fit towards my question. Well with Search to Insight, SAP Analytics Cloud delivers just the same experience. You just type your question, ie.: Show top 5 Products by Country in 2018. It gives me various results ranked by relevance. I can immediately add the resulting chart to my storyboard. Remarkable is that I can ask the question in the langiage of my preference. The embedded ML will understand my question and parse it through the data based on NLP Natural Language Processing technology.
The latest version of Search to Insight now also support speech. It means I can just ask the software my query using the laptop microphone. See below video to see it working.
Smart Discovery 2.0
Smart Discovery is a ML enabled feature that allows to find correlations in the data.
Based on a target metric of choice, Smart Discovery finds correlations, key influencers and data patterns. It also generates a siimulation and creates an overview of outliers in the data . With the recent bettered version Smart Discovery 2.0 the ML algorithms have extended and bring more insights. It now also allows to immediately create a full storyboard at once. One of the ways to apply Smart Discovery 2.0 and have it automatically generate a story for you, is as follows:
click new story
choose Run a Smart Discovery
choose the Model that holds your data
choose the Target Metric you want to Analyse
option: choose filtering or exlusions
Below a video where I show you both Smart Features starting with Search to Insight and later Smart Discovery 2.0.
Lead, guide and make value
Smart Features work a little bit like in below video: the software "leads" and "guides" the user to the best insight of value. The video is on Pool playing, yet the principle of using ML is the same. As a user one is "guided to the solution that brings the best value"; in this case that's making the pot.