day_name day hour risk grade count Friday 1 0 A2 6 Friday 1 0 A3 6 Friday 1 0 D1 3 Friday 1 0 D2 2 Friday 1 0 DN 1 Friday 1 0 DS 6 Friday 1 1 A2 4 Friday 1 1 A3 4 Friday 1 1 D1 1 Friday 1 1 DS 1 Friday 1 2 A3 1     Hi, my data consists of hourly application count for each risk grade by day of the month and day of the week. Attached data here. As the application processors receive these volume of applications and make decisions on them, they are being evaluated by the approval rates etc. Approval rates on A1 are the highest and in general approval rates are higher for A (A1,A2,A3) grade applications. So if I am trying to evaluate application processors performance on approval rates, to be on the fair side, I will have to test first if they receive same frequency of risk grade applciations. For example, if I see higher approval rates for night shift application processos, i need to test to see if they receive more A grades compared to the day shift processors, similarly for weekday processors vs weekend processors, and internet applications vs phone applications etc. To visualize the relation between risk grade volume and hour, I can do a simple 2*2 freq matrix and see if the distribution manually, but as the dimensions/slices to the data increase, riskgrade*day_of_week*hour or riskgrade*day_of_month*day_of_week*hour etc.. if becomes cumbersome to look at the data manually. Is there any procedure that I could use to solve this? I have tried using PROC CORRESP, but not sure if that is the appropriate procedure to use here. Please suggest me which statistical method or procedure to find the association between these variables. Thanks. ... View more

Hi,   I have following dataset. data test; input id date yymmdd8. X Y; format date date9.; datalines; 1 20160131 1 1 1 20160229 1 2 1 20160331 2 3 1 20160430 3 4 1 20160531 3 5 1 20160630 3 6 2 20160131 1 1 2 20160229 1 2 2 20160331 2 3 2 20160430 2 4 2 20160531 3 5 2 20160630 3 6 3 20160131 1 1 3 20160229 4 2 3 20160331 9 3 3 20160430 9 4 ; run;   id date X Y 1 31Jan2016 1 1 1 29Feb2016 1 2 1 31Mar2016 2 3 1 30Apr2016 3 4 1 31May2016 3 5 1 30Jun2016 3 6 2 31Jan2016 1 1 2 29Feb2016 1 2 2 31Mar2016 2 3 2 30Apr2016 2 4 2 31May2016 3 5 2 30Jun2016 3 6 3 31Jan2016 1 1 3 29Feb2016 4 2 3 31Mar2016 9 3 3 30Apr2016 9 4   From this I need to select the record for each ID, at the instance when (X > 2) for the first time.   So from the above dataset my output should be:   id date X Y 1 30Apr2016 3 4 2 31May2016 3 5 3 29Feb2016 4 2     I tried doing: data test; input id date yymmdd8. X Y; format date date9.; datalines; 1 20160131 1 1 1 20160229 1 2 1 20160331 2 3 1 20160430 3 4 1 20160531 3 5 1 20160630 3 6 2 20160131 1 1 2 20160229 1 2 2 20160331 2 3 2 20160430 2 4 2 20160531 3 5 2 20160630 3 6 3 20160131 1 1 3 20160229 4 2 3 20160331 9 3 3 20160430 9 4 ; run; proc sort data=test out=test1; by id descending x date; run; data test2; set test1; by id; if first.id then output; run;   But this would give me: id date X Y 1 30Apr2016 3 4 2 31May2016 3 5 3 31Mar2016 9 3 Wrong entry for Id=3.   Any easy way to do this on a dataset with 60million records.   Thanks   ... View more

Hi , I was working with this before and this has been answered in the post https://communities.sas.com/t5/SAS-Procedures/Create-frequency-count-crosstab-matrix-and-for-each-combination/m-p/302505#M60713     I have a following dataset   ID   attribute1   attribute2   attribute3     BAD (Y-> yes) 1        1               0               0                 Y 2        1               1               0                 N 3        1               0               1                 Y 4        0               1               1                 N 5        1               1               0                 Y 6        1               1               0                 N 7        1               0               1                 Y 8        1               0               1                 N   I want to create frequency count matrix, for each combination of attribute tripped (atrribute = 1 means tripped) like following:                             attribute1    attribute2    attribute3 attribute1                  _                3                  3 attribute2                 3                 _                 1 attribute3                 3                 1                  _   It represents counts of attributes tripped together. Once i have the counts I need to calcualte %BAD in each combination like below.                         attribute1     attribute2     attribute3 attribute1             _               12.50%         25% attribute2        12.50%             _                   0% attribute3            25%              0%                _   it represents out of 3 ID's for the combination atribute1 and atribute2 tripped, one ID went BAD -> so 12.5% BAD And out of 3 ID's for the combination attribute1 and attribute1 tripped, two went BAD -> so 25% BAD And the 1 ID for the combination attribute2 and attribute3 tripped, didnt went BAD -> so 0% BAD       I need to expand this analysis t0 3 variable combinations to get more meaningful results.   My dataset look like:   ID   grade type score attribute1  attribute2 attribute3 attribute4 attribute5 attribute6 BAD (Y-yes/N- No) 1        A 1 100-110 1               0               0                 1               0               0                 Y 2        A 1 120-130 1               1               0                 1               1               0                 N 3        B 2 70-80 1               0               1                 1               0               1                 Y 4        C 3 30-40 0               1               1                 0               1               1                 N 5        A 3 40-50 1               1                0                 1               1                0                 Y 6        B 4 80-90 1               1               0                 1               1               0                  N 7        C 2 40-50 1               0               1                 1               0               1                  Y 8        C 4 30-40 1               0               1                 1               0               1                  N     Im trying to get to this analysis: I am showing values only for grade A as example:           frequency Attributes/Attribute combinations               Attribute1 Attribute2 Attribute4 Attr1&Attr4 Attr1&Attr2 Attr1&Attr2&Attr4 BAD grade score             yes A 100-110 1   1 1         110-120                 120-130               B 90-100                 80-90                 70-80               C 40-50                 30-40                               No A 100-110 2 1 2 2 1 1     110-120                 120-130               B 90-100                 80-90                 70-80               C 40-50                 30-40               I was looking at http://www.datavis.ca/papers/sugi/mdarray/mdarray.pdf on mutildimensional arrays and mosaic displays. Wasn't quite sure how to get to such kind of analysis with this type of data. Any suggestions?     Thanks ... View more