An IoT Future: Analytics Heaven or Hell?
Published on 22 April 16
Mario Lewis
0
While we continue to wait for the IoT explosion that we were always told is about to happen, I can’t help wondering whether it’s really going to give businesses the kind of raw data that would make data scientists jump for joy.
One of the issues that usually pops up in IoT discussions is security. Hand in hand with that is privacy, and there are others as well.
Will large data volumes be one of them?
Its very likely that a lot of IoT use cases will centre around customer relationships and getting closer to customers, giving them more personalized offerings, offering them better convenience and such. But will it really work like that?
I’ve already talked about the possibilities of using telematics to personalize insurance policy pricing in an earlier post. Can the way that IoT setup is designed be used in many more cases. Perhaps it can. But here’s what I’d be concerned about.
We are already deluged with too much information. Email and social media are useful, but they also attract an information overload. Mobile apps already add to it all by sending us notifications and alerts, and some apps give users the option to turn them off.
When IoT applications grow more common maybe many application device owners will turn a lot of them (or their sensors) off too, because of privacy concerns, or because of concerns about having to deal with too much information.
What are the implications of that for analytics? It could go three ways, in my opinion.
Too little data: If too many users turn off the data stream maybe there’d be too little data to analyse, or the analysis would be misleading due to the absence of significant variables.
Just the right amount of data: Enough users leave enough data streams turned on, and there’s just the right amount of variety in the data for useful analytics insights to be harvested.
Too much data: We might be flooded with too much data of too many types. One of the potential risks with having too many varieties of data, and too many data points available is that we might run into an overfit situation, or the setting up of a correlations chain that isn’t really correct from a business interpretation point of view.Somehow the just right scenario seems like a difficult one just now. I think the first and third scenarios are more likely. But the way these things evolve and progress I just may be wrong about that. We’ll just have to wait and see how things pan out. It'd be interesting to know what experts in this field think is going to happen.
This blog is listed under
Data & Information Management
and Networks & IT Infrastructure
Community
Related Posts:
Post a Comment
