Question

I have large table, with around 200 fields. Around a 100 of those fields are to be mapped to one field when creating a dimension table from it. The problem is I have to add the values of those 100 fields one-by-one through like a 100 insert statements. Is there like a loop or something with which i can achieve this more efficiently?

Here’s an example of the code:

    insert into DimTableA(visit_no, patient, facility, icd9, icd9_flag, ip_op)
    select D.registration, D3.med_number, D3.Hosp_Id, D.final_diagnosis_18,'d',IF(D3.Admit_Type in(1,2,3),'Inpatient','Outpatient'),
    from svch_dischs3_s D3, svch_diags_s D

    insert into DimTableA(visit_no, patient, facility, icd9, icd9_flag, ip_op)
    select D.registration, D3.med_number, D3.Hosp_Id, D.final_diagnosis_19,'d',IF(D3.Admit_Type in(1,2,3),'Inpatient','Outpatient'),
    from svch_dischs3_s D3, svch_diags_s D

    insert into DimTableA(visit_no, patient, facility, icd9, icd9_flag, ip_op)
    select D.registration, D3.med_number, D3.Hosp_Id, D.final_diagnosis_20,'d',IF(D3.Admit_Type in(1,2,3),'Inpatient','Outpatient'),
    from svch_dischs3_s D3, svch_diags_s D;

……… and so on

The only field name that changes is the ‘icd9’ input( i.e D.final_diagnosis_18, final_diagnosis_19, final_diagnosis_20 ….)

Any help would be really appreciated, lads. 🙂

Answer 1

You could unpivot those columns using a cross join to a fixed set of numbers like this:

insert into DimTableA (
  visit_no,
  patient,
  facility,
  icd9,
  icd9_flag,
  ip_op
)
select
  D.registration,
  D3.med_number,
  D3.Hosp_Id,
  case N.number
    when 1 then D.final_diagnosis_18
    when 2 then D.final_diagnosis_19
    when 3 then D.final_diagnosis_20
    ...
  end,
  'd',
  IF(D3.Admit_Type in(1,2,3),'Inpatient','Outpatient')
from
  svch_dischs3_s D3,
  svch_diags_s D,
  (
    select 1 as number union all
    select 2           union all
    select 3           union all
    ...  /* up to the number of columns to unpivot */
  ) N

You could create and populate a persistent numbers table instead of the inline view and use the necessary subset of that table in your query. In that case the query would change like this

...
from
  svch_dischs3_s D3,
  svch_diags_s D,
  numbers N
where N.number between 1 and … /* the number of columns to unpivot */