Data quality – garbage in, recycled goodness out

Data quality is a way of life – it’s something that needs monitoring, that needs reporting on…but you need the right tools.

Obviously Sabisu can handle the reporting side, but we’re often asked, “How do you ensure data quality?”

Often this is followed by the old chestnut; “Garbage in…garbage out.”

Well, sometimes that’s the case…but not all the time. Here’s 3 ways Sabisu helps ensure data quality.

Garbage in, recycled goodness out

Sabisu has a suite of algorithms that can clean process data up; up-sampling, interpolation, noise reduction, transient analysis, anomaly detection, cluster analysis…the list goes on.

For example, just in terms of downsampling Sabisu has 4 different algorithms; again, 4 different denoising approaches are available depending on the nature of the signal.

(Most of the time Sabisu will simply choose the right one and use it.)

Unstructured data can be cleaned up too and we’re working on using machine learning (e.g., text mining and semantic algorithms) to process unstructured data further.

Virtuous circles

“If you want to improve something, measure it.”

We hear this a lot and it’s almost true, if you want to change something, measuring it will do that. (There’s a quantum physics relationship there, but i’ll let you find it.)

Using Sabisu to measure process, operational or project performance will improve data quality as the data is needed, so there is a reason to make it accurate. If the responsible person knows the data is important, they seem to take more care over putting it together, or validating it.

To that end we have features like Publication Approvals, which allow users to pause updates while data is checked and validated.

New data, new contexts

The third way we can improve data quality is to capture new forms of data, or data from new places, or data from source.

A great example of this is Plant Mobile, which:

  • eliminates double entry of data, e.g, no need to write down a reading take it back to the control room and enter it into a spreadsheet / SAP interface
  • allows entry only of valid data, e.g., users have to identify the correct asset from a list
  • captures rich metadata to accompany processes, e.g., photographs to go with a plant inspection.


Sabisu Getac Logistics

Another example is where an MS Excel spreadsheet is replaced with a direct connection to process data – which might need denoising or interpolating but at least is trustworthy.

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