3 Replies Latest reply on Jan 27, 2019 10:07 AM by Shinichiro Murakami

# Correlation Matrix Between Two Different Dimensions

I've been working on figuring out the correlation matrix for a few hours now and something seems pretty clear to me... most tutorials will only show you how to compare 1 dimension with any measure.

In my scenario I would like to compare 2 dimensions with one measure.

Dimensions used:

[Model Number]

[Work Center]

Measure used:

[Number of Defects]

I'm trying to find a correlation between the model number and where a defect occurs on the assembly line.

i.e. When I'm building Model X which work centers are more likely to cause a defect?

Defects are characterized (in my data) as event numbers.  There can be 100's of Events (defects) written on the assembly line each day.

As shown within the tutorials I've watched, I add the same table twice in the data source and did an inner join on the [Event No] and ran through the calculation that each tutorial shows you... but this gets me no where.

The tutorials all have the same dimensions (i.e. Model & Model1) on the X and Y axis.

I need my Model numbers on the Y axis and my Work Centers on the X axis.  Then I need my Pearson Corr Calc to be added to the color card to show the correlations.

Here's the calc per the tutorial:

Does anyone have any ideas as to how to go about doing this?

• ###### 1. Re: Correlation Matrix Between Two Different Dimensions

Hey Stephen,

By any chance, could you share your workbook or a mockup of your data as that way would be easier to understand. Also, the calculation you uploaded works with what is displayed in your worksheet, so by itself does not provide much information.

Best,

Diego

1 of 1 people found this helpful
• ###### 2. Re: Correlation Matrix Between Two Different Dimensions

Hi Diego,

I've attached the workbook that I'm referencing.

I also setup sheet1 just as all the tutorials had said to do...  but again, this only allows me to so show a correlation between a single dimension (actually the same one twice per the table it's being pulled from).

I'm beginning to think that in order to compare two different dimensions, I don't need to do a Cartesian Product type of join... I can simply use the Pearson Correlation Coefficient to compare a single table using two different dimensions.... nonetheless, I still have no idea how.

1 of 1 people found this helpful
• ###### 3. Re: Correlation Matrix Between Two Different Dimensions

I am not sure you really seeking "correlation" in this case o not.

I mean I cannot understand which two measure you want to get correlation.

Above simply shows the defect ratio by Area x Model.

You can filter by family or whatever you want.

Thanks,

Shin

1 of 1 people found this helpful