topic using proc corr for subcolumns within columns in SAS Procedures

using proc corr for subcolumns within columns

podarum — Sun, 05 Feb 2012 03:16:00 GMT

Hi, so on to my explanation. I'm trying to use proc corr (on any other correlation procedure) that I can correlated subcolumns within columns. This is a 2 PART question.

Part 1 :The data I have is as below and I'm trying to transpose it so it looks like WANT..

HAVE:

Metric1 Metric2 Metric3

COL1 SCol1 Group1

COL1 SCol2 Group1

COL1 SCol3 Group1

COL1 SCol4 Group1

COL2 SCol5 Group1

COL2 SCol6 Group1

COL2 SCol7 Group1

COL2 SCol8 Group1

COL1 SCol1 Group2

COL1 SCol2 Group2

COL1 SCol3 Group2

COL1 SCol4 Group2

COL2 SCol5 Group2

COL2 SCol6 Group2

COL2 SCol7 Group2

COL2 SCol8 Group2

WANT:

COL1 COL2

SCol1 SCol2 SCol3 Scol4 SCol5 SCol6 SCol7 SCol8

Group1 Metric1

Group1 Metric2

Group1 Metric3

Group2 Metric1

Group2 Metric2

Group2 Metric3

Part 2 of this problem is that I would like to correlate the SCol's within each COL only. So for example only the SCol1-SCol4 would correlate, and same for SCol5-SCol8, as per their respective COL's. thanks

using proc corr for subcolumns within columns

art297 — Sun, 05 Feb 2012 04:01:55 GMT

I don't understand what you are trying to do. Please some example data showing that the data initially looks like, how you want it to look, and what exactly has to be correlated with what.

using proc corr for subcolumns within columns

podarum — Sun, 05 Feb 2012 04:56:30 GMT

Ok, I'll try to simplify it a little:

HAVE:

UN_rt / Price

TO SC Model1 6.7 34.56

TO SC Model2 6.4 56.77

TO NY Model1 6.4 56.77

TO NY Model2 6.4 51.47

TO ET Model1 6.2 12.54

TO ET Model2 6.8 10.54

OTT KN Model1 7.3 37.45

OTT KN Model2 7.3 39.14

OTT VN Model1 7.4 55.76

OTT VN Model2 7.6 58.76

OTT PK Model1 7.1 11.65

OTT PK Model2 7.2 15.67

WANT:

TO OTT

SC NY ET KN VN PK

Model1 UN_rt 6.7 6.4 6.2 7.3 7.4 7.1

Model2 UN_rt 6.4 6.4 6.8 7.3 7.6 7.2

Model1 Price 34.56 56.77 12.54 37.45 55.76 11.65

Model2 Price 56.77 51.47 10.54 39.14 58.76 15.67

Part 2, I want to correlate the subgroups (SC, NY and ET) within TO if there are any corrlations. And same within the OTT group.

As an example, something like :

Group1 (SC+NY)

Group2 (ET)

Group3 (KN+PK)

Group4 (VN)

using proc corr for subcolumns within columns

Reeza — Sun, 05 Feb 2012 05:31:02 GMT

That looks like a transpose rather than correlation? Or is there something that needs to be calculated somwhere that I'm missing.