Correlation on search calculations
Search result calculations can be exported to an Excel file, where they can be analysed further. An interesting workflow is to calculate multiple values for a set of features, to then see how these values correlate. This way we can easily test a hypothesis that two calculations should be correlated. The most straighforward approach is to create a scatter plot in Excel of these calculations, and visually confirm the correlation.
![[en] Batch pressure vs concentration](image/uuid-b1b31d0b-d2ad-4398-2bd0-4c2fe88baccc.png) |
To test our hypothesis that a lower pressure during a batch leads to a lower end concentration of our product, we search for all batches of this year and add the average pressure and the end concentration as search calculations. Exporting to Excel and creating a scatter plot shows most batches fall in a group of high pressure, high correlation, but there is another group with significantly lower pressure and concentration. We could say our hypothesis is confirmed, though we should note that within one group, there seems to be no further correlation between pressure and concentration.
Tipp
You can also plot the duration of a search result against a correlation. An example would be to test if you can operate our equipment longer (result duration) if there was less variation in the pressure (standard deviation calculation). Also, simply plotting a calculated value against the date could nicely display trends in your process over time.
Tipp
You can perform calculations on top of the search results in Excel, and correlate with these. An example would be to divide a delta calculation by the result duration, to get an average rate for the results.