Sunday, January 3, 2016

On data mining

So there was this time then I lived in equatorial Africa and I had near-real-time access (3-5 minutes delay) to all metadata on the biggest MNO (Mobile Network Operator) in Kenya. I used it to filter out all sorts of useful info from those gigabytes of transaction tickets. I was able to trace trends, see network failures, even caught a fraud or two.

This led to my interest in the whole data mining concept and I found this US company which, I still believe, would be the most amazing place to work at. No, not Google. The company I have in mind is Quid Inc. Please see this TED talk. From very simple, easily available info they can reach such a far-reaching conclusions. I can't even imagine how much intelligence I could get out of my Safaricom metadata using this approach. Unfortunately - ,as I wrote in my book - the Marketing Dept. at Safaricom was not able to comprehend this, or they were just too lazy to bother...

Still, I was following (out of sheer curiosity) the world trends in data mining. Then big data was a big world. Now at IBM everything is cognitive, employing the intelligence to filter out the noise and get something meaningful out of this flood of information.

And then... one of the podcasts I subscribe to is SofRep by former US Special Forces operatives, where they just chit-chat while patting themselves how dope they were when they were still operating... And in one of the episodes they bring Mark Lombardi and his data mining and visualizing. Approach much like Quid Inc. but... manually.

Currently going through this book on him and his work. Will let you know what I think once I finish.

No comments: