Support Basic Linear Algebra (Vectors and Matrices)

[mkortink]

I am on a forecasting project where we convert one vector of forecasts into another vector of forecasts by multiplying by a conversion matrix. This is very clumsy and fragile to do in Alteryx meaning we have to drop out to Excel. The ability to do very simple matrix multiplication in Alteryx would be very useful here and in other use cases. I realise you can probably exit to R and do the job, but for something so basic that shouldn't be required.

The relational representation of an mxp matrix is a three column table of cardinality mxp with columns { I , J , A }, where I labels the first index set with index i, J labels the second index set with index j, and A labels the numeric values with value a(i,j).  Given a second pxn  matrix { J, K, B } in relational form we should be able to multiply them to get a mxn matrix { I, K, C} in relational form where of course c(i,k) = sum over j in J of a(i,j)*b(j,k).

Vectors can of course be represented as 1x and x1 matrices. If you really wanted to go to town this could be generalised to array processing ala APL2.

[micjon]

I would also love to see this feature in Alteryx

[mic_jon]

 This functionality could be very useful.

[ShowMeTheData]

I agree, the ability to do matrix multiplication would be extremely useful.  One of my key statistical analysis requires basis matrix manipulation.  To conduct this analysis I have to use another tool.

[DataG8]

Finally found the topic I was looking for this week. In the weekly challenge #168, I was looking for this matrix function as the challenge discusses dice roll probabilities and my first thought was to use probability matrices to model the long run equilibrium. The challenge was much simpler than that so brute forcing all combinations worked. I would love to see some matrix computation tools in Alteryx but the python tool will work for now.