Question

In Excel, I have 10 columns of data from column A to column J, and each column has 1000 rows from row 1 to row 1000. I wonder how to compute the 10 x 10 covariance matrix of the 10 columns of data in Excel?

My partial solution is based on dollar signs and copying formulas:

First I input in a cell a formula =covar($A1:$A1000,A1:A1000).

Then I copy and paste the formula to the cells on the right of the first cell, which gives me =covar($A1:$A1000,B1:B1000) … =covar($A1:$A1000,J1:J1000).

Now I don’t know how I can get =covar(B1:B1000,A1:A1000) … =covar(J1:J1000,A1:A1000), because if I copy and paste the formula to the cells below the first cell, I will get =covar($A2:$A1001,A2:A1001), …, =covar($A1000:$A2001,A1000:A2001) instead.

Thanks!

Answer 1

To make the formula “copy-proof” you can make use of the =OFFSET() function in combination with row and column indices. Example:

• in L1…U1 enter numbers 1, 2, 3, … 10
• in K2…K11 enter numbers 1, 2, 3, … 10
• now copy-proof references to one of the 10 columns A…J. This can be obtained by:
  • =OFFSET($A$1:$A$1000,0,L$1-1) to follow the horizontal index
  • =OFFSET($A$1:$A$1000,0,$K2-1) to follow the vertical index

• and finally you combine the 2 above into

  =COVAR(OFFSET($A$1:$A$1000,0,L$1-1),OFFSET($A$1:$A$1000,0,$K2-1))
  

• this formula that you enter in L2, copy into L2..U11 to obtain your 10×10 matrix

Hope that helps