In the past week, I had a conversation with my ex colleague, who working with Business Intelligence products. For me, it was interesting to hear his opinion regarding the trends, because his focus is mainly on SAP area. As we discussed, in the SAP world, the main focus of the HANA platform, to have real time analytics, and faster result of spreadsheets. My initial thoughts was that I want to challenge, the trends.
The main question which I'm searching:
"What is the main objective of these reports?"
If I remember what I heard previously, the main objective at a report generation was a spreadsheet, which contains large amount of data, with ten thousand of lines, and hundreds of columns. And they will able to form any kind of smaller reports, to show objective information, which give enough information to make decisions.
From my point of view, if I am thinking about Big Data, and Machine learning, an intuitive system like the Wolfram Alpha system, which understood sentences, and give answers based on the information.
After a manager as a question, the machine will find an answer. This is a general procedure which can be seen on any regular meeting. After the answer, the second question is always the same: Why?
And in this case we get a much more correct answer. We get a statistical breakdown of decision tree, which show percentages on facts, which was taken into account to find an answer.
I think this way of gathering information, is much more informative, than a bunch of spreadsheets with 8 arial font sizes. In the future, I think this way of usage of Machine Learning will be much more productive, than any other method, which used previously. The traditional spreadsheet generation will only used for mandatory reports, which required by the authorities. And if I see correctly the future of these systems, the Business Intelligence section of every company will much more rely on Big Data systems, instead of the traditional BI systems.
Last year Google made public available their Machine Learning library, which will give huge step to go into this direction.
If anyone interested, more information about this library can be found on the project page at: http://www.tensorflow.org
Also available Microsofts framework : https://github.com/Microsoft/CNTK
I'm looking forward to the future, when will the first pioneer company announce this kind of Business (Machine) Intelligence software.
