looking for some examples on the use of multidimensional dataelements

Abyot_Gizaw · 8 November 2011 09:41

Hi all,

Can any one provide me an example, possibly from your implementation experience, on the use of multidimensional dataelement where there exists a possibility of using a “dataelement” as dataelement or dimension (option-combination) for example as shown below?

D

E

A

value

B

value

C

Value

value

or

A

B

C

D

value

E

value

Thank you
Abyot.

jason · 8 November 2011 12:38

Hi Abyot,

Not sure if I understand your question, but this is a very typical one for many HIV related data elements.

Male

Female

<1

value

1-15

value

15+

Value

value

However, I am not sure what the difference between this is?

<1

1-15

15+

Male

value

Female

value

Regards,

Jason

···

On Tue, Nov 8, 2011 at 11:41 AM, Abyot Gizaw abyota@gmail.com wrote:

Hi all,

Can any one provide me an example, possibly from your implementation experience, on the use of multidimensional dataelement where there exists a possibility of using a “dataelement” as dataelement or dimension (option-combination) for example as shown below?

D

E

A

value

B

value

C

Value

value

or

A

B

C

D

value

E

value

Thank you
Abyot.

Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

adolphe · 8 November 2011 14:01

Hi
Can this help?

Diagnosis

Under 5 years

5 yrs and above

M

F

M

F

Diarrhoea with Dehydration

Diarrhoea bloody

Diarrhoea Chronic

Intestinal parasites

Malaria confirmed simple

Malaria confirmed simple, in pregnancy

Regards,
Adolphe

···

On 8 November 2011 14:38, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Abyot,
Not sure if I understand your question, but this is a very typical one for many HIV related data elements.

Male

Female

<1

value

1-15

value

15+

Value

value

However, I am not sure what the difference between this is?

<1

1-15

15+

Male

value

Female

value

Regards,

Jason

On Tue, Nov 8, 2011 at 11:41 AM, Abyot Gizaw abyota@gmail.com wrote:

Hi all,

Can any one provide me an example, possibly from your implementation experience, on the use of multidimensional dataelement where there exists a possibility of using a “dataelement” as dataelement or dimension (option-combination) for example as shown below?

D

E

A

value

B

value

C

Value

value

or

A

B

C

D

value

E

value

Thank you
Abyot.

Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

Abyot_Gizaw · 8 November 2011 17:17

Thank you Jason and Kamugunga!

I just wanted to see how the multidimensional is used. Because there are some tricky cases where one can design a particular data collection form as

D

E

A

value

B

value

C

Value

value

and others as

A

B

C

D

value

E

value

so what I am interested to know is what are the merits and demerits between the two? Is this because the model (I mean the multidimensional design) is flexible or confusing one that that kept everything open?

Thank you

Abyot.

···

On Tue, Nov 8, 2011 at 3:01 PM, Kamugunga Adolphe kaadol@gmail.com wrote:

Hi
Can this help?
Diagnosis

Under 5 years

5 yrs and above

M

F

M

F

Diarrhoea with Dehydration

Diarrhoea bloody

Diarrhoea Chronic

Intestinal parasites

Malaria confirmed simple

Malaria confirmed simple, in pregnancy

Regards,
Adolphe

On 8 November 2011 14:38, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Abyot,
Not sure if I understand your question, but this is a very typical one for many HIV related data elements.

Male

Female

<1

value

1-15

value

15+

Value

value

However, I am not sure what the difference between this is?

<1

1-15

15+

Male

value

Female

value

Regards,

Jason

On Tue, Nov 8, 2011 at 11:41 AM, Abyot Gizaw abyota@gmail.com wrote:

Hi all,

Can any one provide me an example, possibly from your implementation experience, on the use of multidimensional dataelement where there exists a possibility of using a “dataelement” as dataelement or dimension (option-combination) for example as shown below?

D

E

A

value

B

value

C

Value

value

or

A

B

C

D

value

E

value

Thank you
Abyot.

Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

jason · 8 November 2011 18:00

As far as I can see there is no difference other than a visual element of presentation. The current model has no sense of hierarchy, but is merely the Cartesian product of possible category options. What you present here seems to me to be completely equivalent in DHIS2 unless I am misinterpreting the model, which is very possible.

···

On Nov 8, 2011 7:17 PM, “Abyot Gizaw” abyota@gmail.com wrote:

Thank you Jason and Kamugunga!

I just wanted to see how the multidimensional is used. Because there are some tricky cases where one can design a particular data collection form as

D

E

A

value

B

value

C

Value

value

and others as

A

B

C

D

value

E

value

so what I am interested to know is what are the merits and demerits between the two? Is this because the model (I mean the multidimensional design) is flexible or confusing one that that kept everything open?

Thank you

Abyot.

On Tue, Nov 8, 2011 at 3:01 PM, Kamugunga Adolphe kaadol@gmail.com wrote:

Hi
Can this help?
Diagnosis

Under 5 years

5 yrs and above

M

F

M

F

Diarrhoea with Dehydration

Diarrhoea bloody

Diarrhoea Chronic

Intestinal parasites

Malaria confirmed simple

Malaria confirmed simple, in pregnancy

Regards,
Adolphe

On 8 November 2011 14:38, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Abyot,
Not sure if I understand your question, but this is a very typical one for many HIV related data elements.

Male

Female

<1

value

1-15

value

15+

Value

value

However, I am not sure what the difference between this is?

<1

1-15

15+

Male

value

Female

value

Regards,

Jason

On Tue, Nov 8, 2011 at 11:41 AM, Abyot Gizaw abyota@gmail.com wrote:

Hi all,

Can any one provide me an example, possibly from your implementation experience, on the use of multidimensional dataelement where there exists a possibility of using a “dataelement” as dataelement or dimension (option-combination) for example as shown below?

D

E

A

value

B

value

C

Value

value

or

A

B

C

D

value

E

value

Thank you
Abyot.

Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

Abyot_Gizaw · 9 November 2011 08:12

As far as I can see there is no difference other than a visual element of presentation.

Yes with the example we mentioned there might not be a difference if we are only interested in the value. But I thought the two are different in the sense that those in the row are dataelements and those in the column are optioncombinations. I thought what we choose to be a dataelement and a dimension affects our subsequent data analysis including indicator definition. I remember encountering a case previously, but I couldn’t remember it.

The current model has no sense of hierarchy, but is merely the Cartesian product of possible category options.

That is true. Currently we are using only the final product (or the last level in the hierarchy) which is the option combination. But the whole idea with the multidimensional is to kind of slice the aggregate figure from different perspectives and corresponding units. Given a total figure of 100, say for example those who have TB, how many of them are male, how many are female. Out of those who are male, how many of them fall under a particular age category, similarly for those who are female…

But we are not utilizing the full potential of dimension and unit conceptualization of the multidimensional. We only use the final optioncombination. I always think the possibility of allowing data collection and analysis at any category or option level - not only optioncombination. But a problem I have is, if we allow that it will really be a mess with lots of overlaps !

···

On Tue, Nov 8, 2011 at 7:00 PM, Jason Pickering jason.p.pickering@gmail.com wrote:

What you present here seems to me to be completely equivalent in DHIS2 unless I am misinterpreting the model, which is very possible.

On Nov 8, 2011 7:17 PM, “Abyot Gizaw” abyota@gmail.com wrote:

Thank you Jason and Kamugunga!

I just wanted to see how the multidimensional is used. Because there are some tricky cases where one can design a particular data collection form as

D

E

A

value

B

value

C

Value

value

and others as

A

B

C

D

value

E

value

so what I am interested to know is what are the merits and demerits between the two? Is this because the model (I mean the multidimensional design) is flexible or confusing one that that kept everything open?

Thank you

Abyot.

On Tue, Nov 8, 2011 at 3:01 PM, Kamugunga Adolphe kaadol@gmail.com wrote:

Hi
Can this help?
Diagnosis

Under 5 years

5 yrs and above

M

F

M

F

Diarrhoea with Dehydration

Diarrhoea bloody

Diarrhoea Chronic

Intestinal parasites

Malaria confirmed simple

Malaria confirmed simple, in pregnancy

Regards,
Adolphe

On 8 November 2011 14:38, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Abyot,
Not sure if I understand your question, but this is a very typical one for many HIV related data elements.

Male

Female

<1

value

1-15

value

15+

Value

value

However, I am not sure what the difference between this is?

<1

1-15

15+

Male

value

Female

value

Regards,

Jason

On Tue, Nov 8, 2011 at 11:41 AM, Abyot Gizaw abyota@gmail.com wrote:

Hi all,

Can any one provide me an example, possibly from your implementation experience, on the use of multidimensional dataelement where there exists a possibility of using a “dataelement” as dataelement or dimension (option-combination) for example as shown below?

D

E

A

value

B

value

C

Value

value

or

A

B

C

D

value

E

value

Thank you
Abyot.

Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

Murod_Latifov · 9 November 2011 08:54

Hi Abyot,

Very interesting, but it seems to me more of levels or categories of the same dimention, not a multidimentionality. Combination of data elements and categories are instances of the same dimention or levels of hierarchy as it comes in data warehouse cube definition. Moreover its representation in datavalue table (fact table) with two keys is ambigious. DHIS2 has the following dimentions: data elements, periods and organisation units and data could be analysed from different angles of these dimentions.

regards,

murod

···

On Wed, Nov 9, 2011 at 1:12 PM, Abyot Gizaw abyota@gmail.com wrote:

On Tue, Nov 8, 2011 at 7:00 PM, Jason Pickering jason.p.pickering@gmail.com wrote:

As far as I can see there is no difference other than a visual element of presentation.

Yes with the example we mentioned there might not be a difference if we are only interested in the value. But I thought the two are different in the sense that those in the row are dataelements and those in the column are optioncombinations. I thought what we choose to be a dataelement and a dimension affects our subsequent data analysis including indicator definition. I remember encountering a case previously, but I couldn’t remember it.

The current model has no sense of hierarchy, but is merely the Cartesian product of possible category options.

That is true. Currently we are using only the final product (or the last level in the hierarchy) which is the option combination. But the whole idea with the multidimensional is to kind of slice the aggregate figure from different perspectives and corresponding units. Given a total figure of 100, say for example those who have TB, how many of them are male, how many are female. Out of those who are male, how many of them fall under a particular age category, similarly for those who are female…

But we are not utilizing the full potential of dimension and unit conceptualization of the multidimensional. We only use the final optioncombination. I always think the possibility of allowing data collection and analysis at any category or option level - not only optioncombination. But a problem I have is, if we allow that it will really be a mess with lots of overlaps !

What you present here seems to me to be completely equivalent in DHIS2 unless I am misinterpreting the model, which is very possible.

On Nov 8, 2011 7:17 PM, “Abyot Gizaw” abyota@gmail.com wrote:

Thank you Jason and Kamugunga!

I just wanted to see how the multidimensional is used. Because there are some tricky cases where one can design a particular data collection form as

D

E

A

value

B

value

C

Value

value

and others as

A

B

C

D

value

E

value

so what I am interested to know is what are the merits and demerits between the two? Is this because the model (I mean the multidimensional design) is flexible or confusing one that that kept everything open?

Thank you

Abyot.

On Tue, Nov 8, 2011 at 3:01 PM, Kamugunga Adolphe kaadol@gmail.com wrote:

Hi
Can this help?
Diagnosis

Under 5 years

5 yrs and above

M

F

M

F

Diarrhoea with Dehydration

Diarrhoea bloody

Diarrhoea Chronic

Intestinal parasites

Malaria confirmed simple

Malaria confirmed simple, in pregnancy

Regards,
Adolphe

On 8 November 2011 14:38, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Abyot,
Not sure if I understand your question, but this is a very typical one for many HIV related data elements.

Male

Female

<1

value

1-15

value

15+

Value

value

However, I am not sure what the difference between this is?

<1

1-15

15+

Male

value

Female

value

Regards,

Jason

On Tue, Nov 8, 2011 at 11:41 AM, Abyot Gizaw abyota@gmail.com wrote:

Hi all,

Can any one provide me an example, possibly from your implementation experience, on the use of multidimensional dataelement where there exists a possibility of using a “dataelement” as dataelement or dimension (option-combination) for example as shown below?

D

E

A

value

B

value

C

Value

value

or

A

B

C

D

value

E

value

Thank you
Abyot.

Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

jason · 9 November 2011 09:01

Hi Murod,

I think your analysis is not entirely correct, or perhaps I am misunderstanding you.

The categoryoptioncombo (i.e. Male, <1 ) is in effect a degenerate dimension which is always bound to the data element and the combination of dataelementid and categoryoptioncomboid is essentially a composite key for this dimension.

One could also analyze the fact table on this dimension as well, to pull out all disease instances of Males under 1 year of age, and not only the data elements/periods/orgunits. In fact, analyzing the data element dimension alone usually makes little sense, except in the context of totals.

Regards,

Jason

···

On Wed, Nov 9, 2011 at 10:54 AM, Murod Latifov mlatifov@gmail.com wrote:

Hi Abyot,

Very interesting, but it seems to me more of levels or categories of the same dimention, not a multidimentionality. Combination of data elements and categories are instances of the same dimention or levels of hierarchy as it comes in data warehouse cube definition. Moreover its representation in datavalue table (fact table) with two keys is ambigious. DHIS2 has the following dimentions: data elements, periods and organisation units and data could be analysed from different angles of these dimentions.

regards,

murod

On Wed, Nov 9, 2011 at 1:12 PM, Abyot Gizaw abyota@gmail.com wrote:

On Tue, Nov 8, 2011 at 7:00 PM, Jason Pickering jason.p.pickering@gmail.com wrote:

As far as I can see there is no difference other than a visual element of presentation.

Yes with the example we mentioned there might not be a difference if we are only interested in the value. But I thought the two are different in the sense that those in the row are dataelements and those in the column are optioncombinations. I thought what we choose to be a dataelement and a dimension affects our subsequent data analysis including indicator definition. I remember encountering a case previously, but I couldn’t remember it.

The current model has no sense of hierarchy, but is merely the Cartesian product of possible category options.

That is true. Currently we are using only the final product (or the last level in the hierarchy) which is the option combination. But the whole idea with the multidimensional is to kind of slice the aggregate figure from different perspectives and corresponding units. Given a total figure of 100, say for example those who have TB, how many of them are male, how many are female. Out of those who are male, how many of them fall under a particular age category, similarly for those who are female…

But we are not utilizing the full potential of dimension and unit conceptualization of the multidimensional. We only use the final optioncombination. I always think the possibility of allowing data collection and analysis at any category or option level - not only optioncombination. But a problem I have is, if we allow that it will really be a mess with lots of overlaps !

What you present here seems to me to be completely equivalent in DHIS2 unless I am misinterpreting the model, which is very possible.

On Nov 8, 2011 7:17 PM, “Abyot Gizaw” abyota@gmail.com wrote:

Thank you Jason and Kamugunga!

I just wanted to see how the multidimensional is used. Because there are some tricky cases where one can design a particular data collection form as

D

E

A

value

B

value

C

Value

value

and others as

A

B

C

D

value

E

value

so what I am interested to know is what are the merits and demerits between the two? Is this because the model (I mean the multidimensional design) is flexible or confusing one that that kept everything open?

Thank you

Abyot.

On Tue, Nov 8, 2011 at 3:01 PM, Kamugunga Adolphe kaadol@gmail.com wrote:

Hi
Can this help?
Diagnosis

Under 5 years

5 yrs and above

M

F

M

F

Diarrhoea with Dehydration

Diarrhoea bloody

Diarrhoea Chronic

Intestinal parasites

Malaria confirmed simple

Malaria confirmed simple, in pregnancy

Regards,
Adolphe

On 8 November 2011 14:38, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Abyot,
Not sure if I understand your question, but this is a very typical one for many HIV related data elements.

Male

Female

<1

value

1-15

value

15+

Value

value

However, I am not sure what the difference between this is?

<1

1-15

15+

Male

value

Female

value

Regards,

Jason

On Tue, Nov 8, 2011 at 11:41 AM, Abyot Gizaw abyota@gmail.com wrote:

Hi all,

Can any one provide me an example, possibly from your implementation experience, on the use of multidimensional dataelement where there exists a possibility of using a “dataelement” as dataelement or dimension (option-combination) for example as shown below?

D

E

A

value

B

value

C

Value

value

or

A

B

C

D

value

E

value

Thank you
Abyot.

Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

olatitle · 9 November 2011 09:02

This section in the user manual uses a real example (RH form) and discusses the various alternatives of data element + category representation and the consequences of these.

http://www.dhis2.org/doc/snapshot/en/user/html/ch21s08.html

Ola

···

Ola Hodne Titlestad (Mr)
HISP
Department of Informatics
University of Oslo

Mobile: +47 48069736
Home address: Vetlandsvn. 95B, 0685 Oslo, Norway. Googlemaps link

On 9 November 2011 11:54, Murod Latifov mlatifov@gmail.com wrote:

Hi Abyot,

Very interesting, but it seems to me more of levels or categories of the same dimention, not a multidimentionality. Combination of data elements and categories are instances of the same dimention or levels of hierarchy as it comes in data warehouse cube definition. Moreover its representation in datavalue table (fact table) with two keys is ambigious. DHIS2 has the following dimentions: data elements, periods and organisation units and data could be analysed from different angles of these dimentions.

regards,

murod

On Wed, Nov 9, 2011 at 1:12 PM, Abyot Gizaw abyota@gmail.com wrote:

On Tue, Nov 8, 2011 at 7:00 PM, Jason Pickering jason.p.pickering@gmail.com wrote:

As far as I can see there is no difference other than a visual element of presentation.

Yes with the example we mentioned there might not be a difference if we are only interested in the value. But I thought the two are different in the sense that those in the row are dataelements and those in the column are optioncombinations. I thought what we choose to be a dataelement and a dimension affects our subsequent data analysis including indicator definition. I remember encountering a case previously, but I couldn’t remember it.

The current model has no sense of hierarchy, but is merely the Cartesian product of possible category options.

That is true. Currently we are using only the final product (or the last level in the hierarchy) which is the option combination. But the whole idea with the multidimensional is to kind of slice the aggregate figure from different perspectives and corresponding units. Given a total figure of 100, say for example those who have TB, how many of them are male, how many are female. Out of those who are male, how many of them fall under a particular age category, similarly for those who are female…

But we are not utilizing the full potential of dimension and unit conceptualization of the multidimensional. We only use the final optioncombination. I always think the possibility of allowing data collection and analysis at any category or option level - not only optioncombination. But a problem I have is, if we allow that it will really be a mess with lots of overlaps !

What you present here seems to me to be completely equivalent in DHIS2 unless I am misinterpreting the model, which is very possible.

On Nov 8, 2011 7:17 PM, “Abyot Gizaw” abyota@gmail.com wrote:

Thank you Jason and Kamugunga!

I just wanted to see how the multidimensional is used. Because there are some tricky cases where one can design a particular data collection form as

D

E

A

value

B

value

C

Value

value

and others as

A

B

C

D

value

E

value

so what I am interested to know is what are the merits and demerits between the two? Is this because the model (I mean the multidimensional design) is flexible or confusing one that that kept everything open?

Thank you

Abyot.

On Tue, Nov 8, 2011 at 3:01 PM, Kamugunga Adolphe kaadol@gmail.com wrote:

Hi
Can this help?
Diagnosis

Under 5 years

5 yrs and above

M

F

M

F

Diarrhoea with Dehydration

Diarrhoea bloody

Diarrhoea Chronic

Intestinal parasites

Malaria confirmed simple

Malaria confirmed simple, in pregnancy

Regards,
Adolphe

On 8 November 2011 14:38, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Abyot,
Not sure if I understand your question, but this is a very typical one for many HIV related data elements.

Male

Female

<1

value

1-15

value

15+

Value

value

However, I am not sure what the difference between this is?

<1

1-15

15+

Male

value

Female

value

Regards,

Jason

On Tue, Nov 8, 2011 at 11:41 AM, Abyot Gizaw abyota@gmail.com wrote:

Hi all,

Can any one provide me an example, possibly from your implementation experience, on the use of multidimensional dataelement where there exists a possibility of using a “dataelement” as dataelement or dimension (option-combination) for example as shown below?

D

E

A

value

B

value

C

Value

value

or

A

B

C

D

value

E

value

Thank you
Abyot.

Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

Murod_Latifov · 9 November 2011 09:50

Hi Jason,

Yes, it is obvious that one may need to have more detailed data rather data element itself. This could be achieved on the way I mentioned it earlier. What is important is to store data in its lowest level, in our case categories (atomicity). Look at the following example:

DE1, DE1 (0-5 age) - value 5

DE1, DE1 (6-27 age) - value 1223

DE1, DE1 (28 above age) - value 998

or

DE1, DE1 (0-5 age, Male) - value 5

DE1, DE1 (6-27 age, Male) - value 1223

DE1, DE1 (28 above age, Male) - value 998

DE1, DE1 (0-5 age, Female) - value 5

DE1, DE1 (6-27 age, Female) - value 1223

DE1, DE1 (28 above age, Female) - value 998

now if we want a general dataelement value we may call select *, sum(value) from fact where dataelement=‘DE1’

or if we wnt groupwise we do select *, sum(value) from fact where dataelement=‘DE1’ group by dataelementcategory

Having composite key to represent the same dimention in fact table is violation of data warehouse rules. This granularity or hierarchy of relartions within single dimention should be managed within dimention itself, as levels.

regards,
murod

···

On Wed, Nov 9, 2011 at 2:01 PM, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Murod,

I think your analysis is not entirely correct, or perhaps I am misunderstanding you.

The categoryoptioncombo (i.e. Male, <1 ) is in effect a degenerate dimension which is always bound to the data element and the combination of dataelementid and categoryoptioncomboid is essentially a composite key for this dimension.

One could also analyze the fact table on this dimension as well, to pull out all disease instances of Males under 1 year of age, and not only the data elements/periods/orgunits. In fact, analyzing the data element dimension alone usually makes little sense, except in the context of totals.

Regards,

Jason

On Wed, Nov 9, 2011 at 10:54 AM, Murod Latifov mlatifov@gmail.com wrote:

Hi Abyot,

Very interesting, but it seems to me more of levels or categories of the same dimention, not a multidimentionality. Combination of data elements and categories are instances of the same dimention or levels of hierarchy as it comes in data warehouse cube definition. Moreover its representation in datavalue table (fact table) with two keys is ambigious. DHIS2 has the following dimentions: data elements, periods and organisation units and data could be analysed from different angles of these dimentions.

regards,

murod

On Wed, Nov 9, 2011 at 1:12 PM, Abyot Gizaw abyota@gmail.com wrote:

On Tue, Nov 8, 2011 at 7:00 PM, Jason Pickering jason.p.pickering@gmail.com wrote:

As far as I can see there is no difference other than a visual element of presentation.

Yes with the example we mentioned there might not be a difference if we are only interested in the value. But I thought the two are different in the sense that those in the row are dataelements and those in the column are optioncombinations. I thought what we choose to be a dataelement and a dimension affects our subsequent data analysis including indicator definition. I remember encountering a case previously, but I couldn’t remember it.

The current model has no sense of hierarchy, but is merely the Cartesian product of possible category options.

That is true. Currently we are using only the final product (or the last level in the hierarchy) which is the option combination. But the whole idea with the multidimensional is to kind of slice the aggregate figure from different perspectives and corresponding units. Given a total figure of 100, say for example those who have TB, how many of them are male, how many are female. Out of those who are male, how many of them fall under a particular age category, similarly for those who are female…

But we are not utilizing the full potential of dimension and unit conceptualization of the multidimensional. We only use the final optioncombination. I always think the possibility of allowing data collection and analysis at any category or option level - not only optioncombination. But a problem I have is, if we allow that it will really be a mess with lots of overlaps !

What you present here seems to me to be completely equivalent in DHIS2 unless I am misinterpreting the model, which is very possible.

On Nov 8, 2011 7:17 PM, “Abyot Gizaw” abyota@gmail.com wrote:

Thank you Jason and Kamugunga!

I just wanted to see how the multidimensional is used. Because there are some tricky cases where one can design a particular data collection form as

D

E

A

value

B

value

C

Value

value

and others as

A

B

C

D

value

E

value

so what I am interested to know is what are the merits and demerits between the two? Is this because the model (I mean the multidimensional design) is flexible or confusing one that that kept everything open?

Thank you

Abyot.

On Tue, Nov 8, 2011 at 3:01 PM, Kamugunga Adolphe kaadol@gmail.com wrote:

Hi
Can this help?
Diagnosis

Under 5 years

5 yrs and above

M

F

M

F

Diarrhoea with Dehydration

Diarrhoea bloody

Diarrhoea Chronic

Intestinal parasites

Malaria confirmed simple

Malaria confirmed simple, in pregnancy

Regards,
Adolphe

On 8 November 2011 14:38, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Abyot,
Not sure if I understand your question, but this is a very typical one for many HIV related data elements.

Male

Female

<1

value

1-15

value

15+

Value

value

However, I am not sure what the difference between this is?

<1

1-15

15+

Male

value

Female

value

Regards,

Jason

On Tue, Nov 8, 2011 at 11:41 AM, Abyot Gizaw abyota@gmail.com wrote:

Hi all,

Can any one provide me an example, possibly from your implementation experience, on the use of multidimensional dataelement where there exists a possibility of using a “dataelement” as dataelement or dimension (option-combination) for example as shown below?

D

E

A

value

B

value

C

Value

value

or

A

B

C

D

value

E

value

Thank you
Abyot.

Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

Abyot_Gizaw · 9 November 2011 10:15

I am not sure I understood your case Murod. But I don’t agree with your analysis that we have “more of levels or categories of the same dimention, not a multidimentionality.” For me there is multidimensionality.

Thank you Ola for the example - that was the sort of example I was looking for !

···

On Wed, Nov 9, 2011 at 10:50 AM, Murod Latifov mlatifov@gmail.com wrote:

Hi Jason,

Yes, it is obvious that one may need to have more detailed data rather data element itself. This could be achieved on the way I mentioned it earlier. What is important is to store data in its lowest level, in our case categories (atomicity). Look at the following example:

DE1, DE1 (0-5 age) - value 5

DE1, DE1 (6-27 age) - value 1223

DE1, DE1 (28 above age) - value 998

or

DE1, DE1 (0-5 age, Male) - value 5

DE1, DE1 (6-27 age, Male) - value 1223

DE1, DE1 (28 above age, Male) - value 998

DE1, DE1 (0-5 age, Female) - value 5

DE1, DE1 (6-27 age, Female) - value 1223

DE1, DE1 (28 above age, Female) - value 998

now if we want a general dataelement value we may call select *, sum(value) from fact where dataelement=‘DE1’

or if we wnt groupwise we do select *, sum(value) from fact where dataelement=‘DE1’ group by dataelementcategory

Having composite key to represent the same dimention in fact table is violation of data warehouse rules. This granularity or hierarchy of relartions within single dimention should be managed within dimention itself, as levels.

regards,
murod

On Wed, Nov 9, 2011 at 2:01 PM, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Murod,

I think your analysis is not entirely correct, or perhaps I am misunderstanding you.

The categoryoptioncombo (i.e. Male, <1 ) is in effect a degenerate dimension which is always bound to the data element and the combination of dataelementid and categoryoptioncomboid is essentially a composite key for this dimension.

One could also analyze the fact table on this dimension as well, to pull out all disease instances of Males under 1 year of age, and not only the data elements/periods/orgunits. In fact, analyzing the data element dimension alone usually makes little sense, except in the context of totals.

Regards,

Jason

On Wed, Nov 9, 2011 at 10:54 AM, Murod Latifov mlatifov@gmail.com wrote:

Hi Abyot,

Very interesting, but it seems to me more of levels or categories of the same dimention, not a multidimentionality. Combination of data elements and categories are instances of the same dimention or levels of hierarchy as it comes in data warehouse cube definition. Moreover its representation in datavalue table (fact table) with two keys is ambigious. DHIS2 has the following dimentions: data elements, periods and organisation units and data could be analysed from different angles of these dimentions.

regards,

murod

On Wed, Nov 9, 2011 at 1:12 PM, Abyot Gizaw abyota@gmail.com wrote:

On Tue, Nov 8, 2011 at 7:00 PM, Jason Pickering jason.p.pickering@gmail.com wrote:

As far as I can see there is no difference other than a visual element of presentation.

Yes with the example we mentioned there might not be a difference if we are only interested in the value. But I thought the two are different in the sense that those in the row are dataelements and those in the column are optioncombinations. I thought what we choose to be a dataelement and a dimension affects our subsequent data analysis including indicator definition. I remember encountering a case previously, but I couldn’t remember it.

The current model has no sense of hierarchy, but is merely the Cartesian product of possible category options.

That is true. Currently we are using only the final product (or the last level in the hierarchy) which is the option combination. But the whole idea with the multidimensional is to kind of slice the aggregate figure from different perspectives and corresponding units. Given a total figure of 100, say for example those who have TB, how many of them are male, how many are female. Out of those who are male, how many of them fall under a particular age category, similarly for those who are female…

But we are not utilizing the full potential of dimension and unit conceptualization of the multidimensional. We only use the final optioncombination. I always think the possibility of allowing data collection and analysis at any category or option level - not only optioncombination. But a problem I have is, if we allow that it will really be a mess with lots of overlaps !

What you present here seems to me to be completely equivalent in DHIS2 unless I am misinterpreting the model, which is very possible.

On Nov 8, 2011 7:17 PM, “Abyot Gizaw” abyota@gmail.com wrote:

Thank you Jason and Kamugunga!

I just wanted to see how the multidimensional is used. Because there are some tricky cases where one can design a particular data collection form as

D

E

A

value

B

value

C

Value

value

and others as

A

B

C

D

value

E

value

so what I am interested to know is what are the merits and demerits between the two? Is this because the model (I mean the multidimensional design) is flexible or confusing one that that kept everything open?

Thank you

Abyot.

On Tue, Nov 8, 2011 at 3:01 PM, Kamugunga Adolphe kaadol@gmail.com wrote:

Hi
Can this help?
Diagnosis

Under 5 years

5 yrs and above

M

F

M

F

Diarrhoea with Dehydration

Diarrhoea bloody

Diarrhoea Chronic

Intestinal parasites

Malaria confirmed simple

Malaria confirmed simple, in pregnancy

Regards,
Adolphe

On 8 November 2011 14:38, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Abyot,
Not sure if I understand your question, but this is a very typical one for many HIV related data elements.

Male

Female

<1

value

1-15

value

15+

Value

value

However, I am not sure what the difference between this is?

<1

1-15

15+

Male

value

Female

value

Regards,

Jason

On Tue, Nov 8, 2011 at 11:41 AM, Abyot Gizaw abyota@gmail.com wrote:

Hi all,

Can any one provide me an example, possibly from your implementation experience, on the use of multidimensional dataelement where there exists a possibility of using a “dataelement” as dataelement or dimension (option-combination) for example as shown below?

D

E

A

value

B

value

C

Value

value

or

A

B

C

D

value

E

value

Thank you
Abyot.

Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

Murod_Latifov · 9 November 2011 11:50

Hi Abyot,

My case is that you may not always have ‘age <5’ as dimention instance. It is linked to certain data elements, not to all. From here category cannot be a dimention by itself. You cannot select ‘age <5’ and match it with some other dimentions, like time and period, it will not give you any meaning, but as part of data element it has some meaning. What you are presenting is two dimentional presentation of the same dimention. This twodimentionality within single dimention leads to composite primary key in datavalue table and violates data warehouse design rules. All dimentions should have valid value all the time, levels have validity within dimention they are linked to. In your multidimentionality ‘Hospital beds (age <5)’ will give wrong value, even meaning will be obsured.

regards,
murod

···

On Wed, Nov 9, 2011 at 3:15 PM, Abyot Gizaw abyota@gmail.com wrote:

I am not sure I understood your case Murod. But I don’t agree with your analysis that we have “more of levels or categories of the same dimention, not a multidimentionality.” For me there is multidimensionality.

Thank you Ola for the example - that was the sort of example I was looking for !

On Wed, Nov 9, 2011 at 10:50 AM, Murod Latifov mlatifov@gmail.com wrote:

Hi Jason,

Yes, it is obvious that one may need to have more detailed data rather data element itself. This could be achieved on the way I mentioned it earlier. What is important is to store data in its lowest level, in our case categories (atomicity). Look at the following example:

DE1, DE1 (0-5 age) - value 5

DE1, DE1 (6-27 age) - value 1223

DE1, DE1 (28 above age) - value 998

or

DE1, DE1 (0-5 age, Male) - value 5

DE1, DE1 (6-27 age, Male) - value 1223

DE1, DE1 (28 above age, Male) - value 998

DE1, DE1 (0-5 age, Female) - value 5

DE1, DE1 (6-27 age, Female) - value 1223

DE1, DE1 (28 above age, Female) - value 998

now if we want a general dataelement value we may call select *, sum(value) from fact where dataelement=‘DE1’

or if we wnt groupwise we do select *, sum(value) from fact where dataelement=‘DE1’ group by dataelementcategory

Having composite key to represent the same dimention in fact table is violation of data warehouse rules. This granularity or hierarchy of relartions within single dimention should be managed within dimention itself, as levels.

regards,
murod

On Wed, Nov 9, 2011 at 2:01 PM, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Murod,

I think your analysis is not entirely correct, or perhaps I am misunderstanding you.

The categoryoptioncombo (i.e. Male, <1 ) is in effect a degenerate dimension which is always bound to the data element and the combination of dataelementid and categoryoptioncomboid is essentially a composite key for this dimension.

One could also analyze the fact table on this dimension as well, to pull out all disease instances of Males under 1 year of age, and not only the data elements/periods/orgunits. In fact, analyzing the data element dimension alone usually makes little sense, except in the context of totals.

Regards,

Jason

On Wed, Nov 9, 2011 at 10:54 AM, Murod Latifov mlatifov@gmail.com wrote:

Hi Abyot,

Very interesting, but it seems to me more of levels or categories of the same dimention, not a multidimentionality. Combination of data elements and categories are instances of the same dimention or levels of hierarchy as it comes in data warehouse cube definition. Moreover its representation in datavalue table (fact table) with two keys is ambigious. DHIS2 has the following dimentions: data elements, periods and organisation units and data could be analysed from different angles of these dimentions.

regards,

murod

On Wed, Nov 9, 2011 at 1:12 PM, Abyot Gizaw abyota@gmail.com wrote:

On Tue, Nov 8, 2011 at 7:00 PM, Jason Pickering jason.p.pickering@gmail.com wrote:

As far as I can see there is no difference other than a visual element of presentation.

Yes with the example we mentioned there might not be a difference if we are only interested in the value. But I thought the two are different in the sense that those in the row are dataelements and those in the column are optioncombinations. I thought what we choose to be a dataelement and a dimension affects our subsequent data analysis including indicator definition. I remember encountering a case previously, but I couldn’t remember it.

The current model has no sense of hierarchy, but is merely the Cartesian product of possible category options.

That is true. Currently we are using only the final product (or the last level in the hierarchy) which is the option combination. But the whole idea with the multidimensional is to kind of slice the aggregate figure from different perspectives and corresponding units. Given a total figure of 100, say for example those who have TB, how many of them are male, how many are female. Out of those who are male, how many of them fall under a particular age category, similarly for those who are female…

But we are not utilizing the full potential of dimension and unit conceptualization of the multidimensional. We only use the final optioncombination. I always think the possibility of allowing data collection and analysis at any category or option level - not only optioncombination. But a problem I have is, if we allow that it will really be a mess with lots of overlaps !

What you present here seems to me to be completely equivalent in DHIS2 unless I am misinterpreting the model, which is very possible.

On Nov 8, 2011 7:17 PM, “Abyot Gizaw” abyota@gmail.com wrote:

Thank you Jason and Kamugunga!

I just wanted to see how the multidimensional is used. Because there are some tricky cases where one can design a particular data collection form as

D

E

A

value

B

value

C

Value

value

and others as

A

B

C

D

value

E

value

so what I am interested to know is what are the merits and demerits between the two? Is this because the model (I mean the multidimensional design) is flexible or confusing one that that kept everything open?

Thank you

Abyot.

On Tue, Nov 8, 2011 at 3:01 PM, Kamugunga Adolphe kaadol@gmail.com wrote:

Hi
Can this help?
Diagnosis

Under 5 years

5 yrs and above

M

F

M

F

Diarrhoea with Dehydration

Diarrhoea bloody

Diarrhoea Chronic

Intestinal parasites

Malaria confirmed simple

Malaria confirmed simple, in pregnancy

Regards,
Adolphe

On 8 November 2011 14:38, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Abyot,
Not sure if I understand your question, but this is a very typical one for many HIV related data elements.

Male

Female

<1

value

1-15

value

15+

Value

value

However, I am not sure what the difference between this is?

<1

1-15

15+

Male

value

Female

value

Regards,

Jason

On Tue, Nov 8, 2011 at 11:41 AM, Abyot Gizaw abyota@gmail.com wrote:

Hi all,

Can any one provide me an example, possibly from your implementation experience, on the use of multidimensional dataelement where there exists a possibility of using a “dataelement” as dataelement or dimension (option-combination) for example as shown below?

D

E

A

value

B

value

C

Value

value

or

A

B

C

D

value

E

value

Thank you
Abyot.

Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp