Hello everyone, it has been a while since I posted a blog. We have been working a lot on data quality with our PI System customers. It never ceases to amaze me how overlooked data quality is when people configure their tags and their PI System in general and everyone seems to ignore this. It can cost you big time. It seems like everyone is focused on what they can DO with the data (analytics, machine learning, etc.) and they just ignore whether the data is even any good or not.
We are also seeing a trend in the IIoT space where fast data collection and only on exception (when the data changes) are prevailing ideas. Below, I rant about three examples where these ideas can hurt you instead of help you.
These are examples of problems we have found recently while performing data fidelity studies for our customers that would have caused problems in advanced analytics initiatives; heck, one of them caused visualization of the data to look wrong!!
Here are some recent videos that I have posted on this subject below.
The first one shows two examples where too much data is a bad thing:
This shows something that a lot of people overlook when configuring advise tags:
Here is a little more about the study (with my twisted sense of humor mixed in around a golf analogy):