Simple process prediction with Edge Correlation

We’ve recently added Edge Correlation to our analytics tab in Widget Working View.

This algorithm aligns the series on a chart depending on their features to show you if their behaviour correlates. It allows for lag between parameters, shifting the curves to fit and tell you what the time difference is.

The lag inherent in many processes makes this a tricky thing to determine any other way.

The algorithm returns qualitative information to show the lag (how much the curve has been shifted to fit) and what the confidence level in the correlation is.

This means Edge Correlation also functions as a simple prediction mechanism.

edge-of-correlation

Above is a good example of how Edge Correlation could be used to determine a link between a process parameter (e.g., temperature change) and a reduction in production output.

Using Edge Correlation immediately shows tht temperature change affects production output 3 hours later. It’s as simple as that.

Users can then leave a Note to inform other users, get feedback and instigate remedial or preventative actions.

Sabisu has several correlation detection algorithms it can use but Edge Correlation is the most reliable and easiest to use; if there’s a correlation, it’ll be clear.

Be on the look out for more analytics functions appearing on Widget Working View soon.

You’ll also notce we’ve improved Sabisu Notes, so no matter where you make a note the relevant Community Members will always receive a real-time notification.

As always with Sabisu, it’s not just about finding the story in the data; it’s about sharing it.

Contact us

We’re always interested in hearing from you with any comments or suggestions, feel free to get in touch.

 

Start typing and press Enter to search