BROADCASTS.com

b'

Tom Davenport has a very easy way to focus in on what\'s needed -- see the picture with the article. Divide the needs between 3 time periods: past, present, and future. And divide between known and unknown.

\\n
1. Most products I\'ve seen are good at past and present knowns -- think data visualization DOMO or Tableau. Great reports and Dashboards.
\\n
2. If you want to dig into the unknown past (like why did this happen?), then you\'ll need something like Splunk or other logfile miners. Something to crunch a lot of data and connect the dots, usually by time stamps or other machine data that\'s hard for humans to notice quickly.
\\n
3. Future Knowns are what most people sell as analytics -- think more excel sheet analytics. If this continues like it has been, then this will happen. If we keep driving, we\'ll run out of gas in 23 miles type of analytics. Nice but limited in value.
\\n
4. Unknowns -- this is where all the fun lives. Past? root cause analysis.
\\n
5. Present unknowns? Think retail -- what\'s the best next action? What do I suggest now?
\\n
6. And everyone\'s favorite: Future unknowns -- think predictive and prescriptive. Given a wildly variable future, what\'s likely to happen and what should I do. This is the land of AI -- learn for me and guide me
\\n