SQL Server Management Studio 2012 Pivot/Cross Tab Query - sql-server

Question

I'm using SQL Server Management Studio 2012 and would like to create a pivot/cross tab query for a table with over 2300 rows.
The table has 5 columns:
- name
- group
- status
- date
- count
There are about 580 distinct names.
Each name is associated to 4 different groups (A, B, C, and D).
Each group has a complete status of yes or no.
A date is associated to each status when completed. Otherwise, status is NULL.
The count column is only applicable to group B and D and is an integer value.
SAMPLE A:
name group status date count
A.A.1 A yes 5/23 NULL
A.A.1 B yes 5/27 112
A.A.1 C yes 6/4 NULL
A.A.1 D yes 6/15 122
A.B.2 A yes 5/25 NULL
A.B.2 B yes 6/1 119
A.B.2 C no NULL NULL
A.B.2 D no NULL NULL
I am trying to display the status of each name as the field values across 11 columns :
- name
- group A
- group A date
- group B
- group B date
- group B count
- group C
- group C date
- group D
- group D date
- group D count
The 'name' column would have the 580 distinct names with their corresponding group data across A, B, C, and D.
SAMPLE B:
nm grp_A A_day grp_B B_day B_ct grp_C C_day grp_D D_day D_ct
A.A.1 yes 5/23 yes 5/27 112 yes 6/4 yes 6/15 122
A.B.2 yes 5/25 yes 6/1 119 no NULL no NULL NULL
(column names have been changed to fit into this question section's format)
Ultimately, the result should have all 580 distinct names in the first column and its corresponding status for each group, the date of completion (or NULL if it has not been completed yet), and the count for groups B and D.
I've tried using a CASE statement, but it generates the names once for each group, resulting in the original table being spaced out across the 11 coulmns.
SAMPLE C:
nm grp_A A_day grp_B B_day B_ct grp_C C_day grp_D D_day D_ct
A.A.1 yes 5/23
A.A.1 yes 5/27 112
A.A.1 yes 6/4
A.A.1 yes 6/15 122
A.B.2 yes 5/25
A.B.2 yes 6/1 119
A.B.2 no NULL
A.B.2 no NULL NULL
What am I doing wrong? Please help!
-- K-moj

Answer 1

Without seeing your query I am guessing but if you are trying to PIVOT the data with a CASE expression my suggestion would be to add an aggregate function around the CASE.
select
name,
max(case when [group] = 'A' then status end) grp_A,
max(case when [group] = 'A' then date end) A_day,
max(case when [group] = 'A' then [count] end) A_ct,
max(case when [group] = 'B' then status end) grp_B,
max(case when [group] = 'B' then date end) B_day,
max(case when [group] = 'B' then [count] end) B_ct,
max(case when [group] = 'C' then status end) grp_C,
max(case when [group] = 'C' then date end) C_day,
max(case when [group] = 'C' then [count] end) C_ct,
max(case when [group] = 'D' then status end) grp_D,
max(case when [group] = 'D' then date end) D_day,
max(case when [group] = 'D' then [count] end) D_ct
from yourtable
group by name
See SQL Fiddle with Demo.
If you want to use the PIVOT function then you will need to first look at unpivoting the status, date and count columns first, then pivot them in the the final result.
An UNPIVOT is when you convert multiple columns of data into multiple rows. You can unpivot the status, date and count columns using a variety of methods. Since you are using SQL Server 2012 you can use CROSS APPLY with a VALUES clause. The code to convert the columns into rows will be:
select name,
col = col+'_'+[group],
value
from yourtable
cross apply
(
values
('grp', status),
('day', [date]),
('ct', cast([count] as varchar(10)))
) c(col, value)
See Demo. This gives a result:
| NAME | COL | VALUE |
| A.A.1 | grp_A | yes |
| A.A.1 | day_A | 5/23 |
| A.A.1 | ct_A | (null) |
| A.A.1 | grp_B | yes |
| A.A.1 | day_B | 5/27 |
| A.A.1 | ct_B | 112 |
Instead of having multiple columns that you want to pivot, you now have all values to be turned into new columns in value and the new column names in col. You can then apply the PIVOT function so the full code will be similar to the following:
select name,
grp_A, day_A, ct_A,
grp_B, day_B, ct_B,
grp_C, day_C, ct_C,
grp_D, day_D, ct_D
from
(
select name,
col = col+'_'+[group],
value
from yourtable
cross apply
(
values
('grp', status),
('day', [date]),
('ct', cast([count] as varchar(10)))
) c(col, value)
) d
pivot
(
max(value)
for col in (grp_A, day_A, ct_A,
grp_B, day_B, ct_B,
grp_C, day_C, ct_C,
grp_D, day_D, ct_D)
) piv
See SQL Fiddle with Demo