http://bit.ly/2cPkeqF
How do I get started with embedded analytics?http://scn.sap.com/community/s4hana/blog/2016/05/27/getting-started-with-s4-hana-embedded-analytics
How does the S/4 HANA embedded analytics architecture look like?http://scn.sap.com/community/s4hana/blog/2016/09/27/journey-of-datafrom-tables-to-tiles--a-sneak-peek-into-the-s4-hana-embedded-analytics-architecture
How Embedded Analytics unlocks the value for the Digital Enterprises:
As the digitization imperative transcends from the boardroom and starts to make its way into more mainstream enterprise execution, the key question related to value realization becomes even more pertinent and perplexing. While the “Fear of Missing Out” (FOMO) and “Disruptive Innovation” ensures that every strategy presentation, keynote and roadmap discussion doles out elaborate reference to the “new” world order of digital enterprise, on ground, the business and process owners are having a tough time putting together the business case for the next big bang.
Having worked with both perspectives I intend to present a possible solution to this value dichotomy. While in my previous blog (http://bit.ly/2a2sfcq) I had presented the solution from an economic perspective, in this series I would propose solution via the embedded analytics. We would develop our solution strategy for Digitization based on the 3 most important characteristics namely - velocity, data volumes and variability.
The digitization war though initiated at the boardroom, would be won at the frontlines of the organization. The speed at which the information is accumulated at the frontlines of the organization, leaves no option but to decentralize decision making beyond the walled gardens of the corporate headquarters. Earlier the information could have traversed through the maze of the organizational hierarchy and the response to immediate crisis or a situation could wait till the summarized information reached the highest levels of the corporate echelons.
Adding to this complexity we had separate systems of records namely the transactional systems like ERP and a separate system for analytics primarily the data warehousing and business intelligence solutions. This separation of responsibilities primarily due to the limitation of the technology coupled with the need for centralized decision making was the norm.
The speed of digitization has taken this trend head-on and the companies which would be able to decentralize decision making farthest to the frontlines will emerge as the winners in this new world order. We would need to redesign the human interaction with the traditional enterprise systems where the information received from the system should be highly contextualized and role based. Instead of offering a vanilla transaction to fit one size for all, we would need to empower the frontline with the context and offer them the view of the business in real time in the device of their choice.
Most important the artificial separation between the system of record and system of information must go and we need to provide a single solution for the operational analytics where the empowered frontline organization can react to the insights in real time. This is achieved by the embedded analytics and the figure below provides an example of how such an embedded analytics snapshot could look like for different users.
In the second step we need to augment the information provided to the business user with the view of the business across the organization and not just in silos. This would mean the ability to combine massive amounts of information from the underlying system of records, create a model and run analytics on this, all in real time. For a simple tile which the user views in the device of his choice, in the background we should be able to aggregate the information across several hundred transaction table in real time and be able to provide the “live” view of the business to the enterprise.
Here is an example below where I have taken the snapshot from an actual ERP systems and just outlined few of the several hundred tables which goes in to create a model, based on which the information is then provided to the end user. This concept is critical for the decentralized decision making to succeed because with this process we lay bare in the hands of the end user the view of all the business which he needs to make a decision and in real-time. Be aware that we have not replicated the information in another system thereby preventing latency and the rigidity of modeling for decision making.
Now comes the most crucial part of the entire exercise – the ability to automate machine learning algorithms which could continuously run through the mega models we created in the previous step. As with digitization the business are “always-on” and need to continuously monitor patterns for any outliers it would not be feasible for just a human analyst to keep track of all the variability associated with the digitization. Autonomous predictive machine learning algorithms now lays the foundation for supporting decision making with the help of bots which are continuously monitoring the complex models for any exception and alerts the business user when there is a need for action.
In the figure below you would find an example of the traversal path which the algorithm could monitor, finally ending up at the source of the information (ERP tables or augmented with the information captured from various IOT devices) containing all the relevant transactional information captured by the enterprise.
Nenhum comentário:
Postar um comentário
Observação: somente um membro deste blog pode postar um comentário.