Partial total in load script for graph category

[#1]

I'm trying to develop a graph with the dimension of 'Diab_Hb_line' with left to right being Recorded, <=7%, > 7% & <= 8%, > 8% & < 10%, >= 10%, No HBA1c.

Currently I have a work around as below load script where I pull the 'Recorded' from another dataset.  What I would like instead is to add the <=7%, > 7% & <= 8%, > 8% & < 10%, >= 10% amounts to create a 'Recorded' amount BUT also have them displayed separately on the graph as above.  I can't seem to find load script which has the kind of set analysis for a sum I am looking for.

The reason I'd like this is that there are a few other graphs I need to do a similar thing to and there is no other dataset with the recorded total for these.

Thanks.

     if(SubField(Diab_Ind_Type,'_',2) = 'Total Patients with valid HbA1c percentage', 'Recorded',

      if(Diab_Hbraw = '<= 7%', '<= 7%',

      if(Diab_Hbraw = '> 7% & <= 8%', '> 7% & <= 8%',

      if(Diab_Hbraw = '> 8% & < 10%', '> 8% & < 10%',

      if(Diab_Hbraw = '>= 10%', '>= 10%',

      if(Diab_Hbraw = 'No HBA1c Recorded', 'No HBA1c'

     ))))))as Diab_Hb_line

[swuehl]

If I understood correctly, you could create a link table linked to Diab_Hbraw and basically load your linked field twice, once with the detailed classification and once with all Diab_Hbraw values linked to 'Recorded'.

Similar to what is shown here

Re: Lixt Box with Ranges of Values

[#3]

I don't know what a link table is. Could you provide the load script for your idea please?

[swuehl]

Well, link table might not be the correct phrase, maybe group table would be more appropriate.

There is some script code in the thread I referenced above and a full example in the cookbook sample file that Rob is linking in his answer.

If you still have problems getting your issue resolved, please post a small sample QVW and a detailed description of your requested result.

[#5]

Sorry I don't have an IT background so trying to convert some of the script examples is a real challenge. I've been struggling with this for days and was pretty impressed with my ability to get the work around script.

This is the end graph I want when considering percentages but I also work with numbers as well.

The raw data has the following where each has a number of patients with value in their HbA1c test.  So for example Practice 1 may have 300 people who have a Diab_Hbraw of <=7%, etc as below

      Diab_Hbraw = '<= 7%'                          ---300

      Diab_Hbraw = '> 7% & <= 8%'              ---100

      Diab_Hbraw = '> 8% & < 10%'               --- 200

      Diab_Hbraw = '>= 10%'                         ---50

      Diab_Hbraw = 'No HBA1c Recorded'       ---225

On my graph I would like Recorded (300+100+200+50) so 650; <=7% 300; > 7% & <= 8% 100; > 8% & < 10% 200; >= 10% 50; No HbA1c 225.

I have two datasets combined in this graph - my load script above shows how I pull the 'Recorded' from one dataset and the rest from another. My data is for multiple practices so I select by practice and month.  [That is irrelevant really]  All I'm looking for is the load script to combine 4 of these bars to make a recorded bar so I don't have to use 2 datasets (ie by summing

Diab_Hbraw = '<= 7%'

Diab_Hbraw = '> 7% & <= 8%'

Diab_Hbraw = '> 8% & < 10%'

Diab_Hbraw = '>= 10%'  )

I don't know how to sum in the load script, particularly when needing to use set values as above.

[swuehl]

Clea,

the script may look like this (first LOAD only to recreate your data sample, the grouping takes place in the second and third LOAD):

Patients:
LOAD Diab_Hbraw, RowNo() as RecID
WHILE iterno() <= Num;
LOAD Dual(Diab_Hbraw,recno()) as Diab_Hbraw,  Num INLINE [
Diab_Hbraw,Num 
'<= 7%' ,300
'> 7% & <= 8%',100
'> 8% & < 10%',200
'>= 10%',50
'No HBA1c Recorded',225
];

Group:
LOAD Diab_Hbraw, Diab_Hbraw as Diab_Hbraw_Grouped
Resident Patients
WHERE NOT Exists(Diab_Hbraw_Grouped,Diab_Hbraw);


LOAD Diab_Hbraw, Dual('Recorded',0) as Diab_Hbraw_Grouped
Resident Group
WHERE Diab_Hbraw < 5; // 5 denotes last entry in the list of dual values, all Diab_Hbraw except 'no HBA1c Recorded'

This creates two tables, first is your current fact table with all patient records and the Diab_Hbraw field, the second table is just used to group your Diab_Hbraw values:

Diab_Hbraw	Diab_Hbraw_Grouped
<= 7%	Recorded
> 7% & <= 8%	Recorded
> 8% & < 10%	Recorded
>= 10%	Recorded
<= 7%	<= 7%
> 7% & <= 8%	> 7% & <= 8%
> 8% & < 10%	> 8% & < 10%
>= 10%	>= 10%
No HBA1c Recorded	No HBA1c Recorded

Now you can create your charts, using the Diab_Hbraw_Grouped field as dimension and Count(DISTINCT RecID) as expression (resp. Count(DISTINCT RecID) / Count(DISTINCT TOTAL RecID) for the percentage).

Hope this helps,

Stefan

[#7]

Hi Stefan,

Thank you for this. I hadn't considered this alternative due to not knowing about it. Thanks for broadening my knowledge.  I ended up playing around with my data before I got your response and ended up adding a field in using plus signs - I had tried the sum function but it wasn't working in the load script. I will have a play with your idea but here is my script in case others want to use it too.

Due to my raw data I have to transpose it and then crosstable it.  I just added the extra field in before cross tabling so it would be part of the set.

Diab_Hb_sum:

LOAD

     date(Data_Date ,'MMM-YY') as Date,

     Place,

     [<= 7%]+[> 7% & <= 8%]+[> 8% & < 10%]+[>= 10%] as Recorded,

     [<= 7%],

     [> 7% & <= 8%],

     [> 8% & < 10%],

     [>= 10%],

     [No HBA1c Recorded] as [No HbA1c]

FROM

Source_Data.xls

(biff, embedded labels, table is HbA1craw$, filters(

Remove(Row, Pos(Top, 3)),

Transpose()

))

;

//Remove row 3 as it has a total in it and is not needed for the set

Diab_Hb_yah:

CrossTable(Diab_group, Diab_group_Pop, 2)

LOAD *

Resident Diab_Hb_sum;

DROP Table Diab_Hb_sum;