dhis2-devs team mailing list archive

[Knut Staring <knutst@xxxxxxxxx>]

I agree that there is no sense in trying to offer "everything", which would
be a nightmare to program and confusing to users. And there are clear
advantages to guiding people through paths that will work. However, I don't
like forcing people unnecessarily through only one workflow - that is what
we have today, and it can be very cumbersome. If you have already made a
bunch of selections, it is much more user friendly to not have to make them
again in a completely different screen - and there should be several ways to
get to the concise data display.

Jason, if you have time, you and I could collaborate on a step-by-step
blueprint for this. We need to break suggested improvements into smaller
chunks and not have this become a huge undertaking. It is a good idea to see
how things are done in DevInfo, see what works well and what doesn't - same
for CRIS. Maybe the best place to start when considering how we want data
browsing to end up is the current Report Table definition screen - with the
kind of Web Pivot view just another take on the HTML output? Seems to me
there is both some consolidation of output and linkages to the output
needed?

Knut

2010/1/21 Jason Pickering <jason.p.pickering@xxxxxxxxx>

> I think it would be good to have a look at DevInfo. I thought I would
> never write that, but there you go. They have a sort of wizard, that
> allows people to select different dimensions (time, places,
> indicators) and then choose the output format (tables, maps, graphs).
> It is a very constrained workflow, but it is a defined and flexible
> workflow, and might be worth taking a look at.
>
> Regards,
> Jason
>
>
>
>
> 2010/1/21 Knut Staring <knutst@xxxxxxxxx>:
> > 2010/1/21 Knut Staring <knutst@xxxxxxxxx>
> >>
> >> 2010/1/21 Lars Helge Øverland <larshelge@xxxxxxxxx>
> >>>
> >>> On Wed, Jan 20, 2010 at 10:03 AM, Ola Hodne Titlestad
> >>> <olatitle@xxxxxxxxx> wrote:
> >>>>
> >>>> Hi,
> >>>>
> >>>> I agree that the data browser results can be misleading to an
> untrained
> >>>> eye. A first thing to do is to actually provide a heading above the
> data
> >>>> (count) table that says something like "Number of values reported
> for:". I
> >>>> agree that we can call it something else, and "Data summary" is not a
> bad
> >>>> name. ( I am also not sure the browse by orgunit actually works, at
> least it
> >>>> doesn't seem to aggregate up the counts as most districts have '0'
> values in
> >>>> that view. )
> >>>>
> >>>> I must say that I actually find the data entry screen quite useful for
> >>>> looking up raw data a given orgunit-period-dataset combination, and
> find it
> >>>> a very fast tool to switch between periods, orgunits and datasets for
> such a
> >>>> view. When it comes to other views into the raw data, like a few data
> >>>> elements over many periods or other pivoted views I agree we need
> something
> >>>> more like a web pivot tool.
> >>>>
> >>>> I actually think we should build on and possibly rename the pivot
> table
> >>>> tool in Reports to "Data browser". After all we are not interested in
> >>>> developing a fully fledged pivot tool, and data browser reflects more
> of
> >>>> what that tool is. Let's try to brainstorm what kind of functionality
> we
> >>>> would have to add in order to come up with a proper data browser. Here
> is a
> >>>> few things I can think of right now:
> >>>>
> >>>> - support for data elements, not just indicators
> >>>> - filter by dataset, not only groups
> >>>> - automatic datamart export triggered in the background (like with
> >>>> report tables)
> >>>> - filter (parameter selection window like datamart export etc.) to
> >>>> select any data element, period, orgunit  (at the same level) and then
> get
> >>>> an ad-hoc pivot table from that
> >>>>
> >>>> What else?
> >>>>
> >>>> Ola
> >>>
> >>>
> >>> I must agree a bit with Ola here... The work-flow of such a browser
> would
> >>> be to select orgunit, then dataset, then period and then get some sort
> of
> >>> display of the data - which is exactly what we have in the data entry
> >>> screen.
> >
> > I don't understand why you think this would be the only workflow - in
> fact I
> > think a user would usually want to leave one or two of those core
> dimensions
> > free, and only select one of them (i.e. the what, where, when - called
> the
> > "grain" in data warehouse
> > terminology http://en.wikipedia.org/wiki/Fact_table )
> > To me, a typical use case would be to initially select between either raw
> > datavalues or indicator values (though even this may be too restrictive),
> > then an orgunit level and then a period type. These, but only these are
> > essential.
> > Then,  you would select *either* a dataelement, a period or an orgunit.
> > This would result in three possible matrix views:
> > 1) Orgunits X Periods for one selected dataelement or indicator
> > 2) Orgunits X Dataelements for one selected period
> > 3) Datalements x Periods
> > Ideally, it would be easy to change the selections smoothly, from drop
> down
> > lists or arrows.
> > The openhealth prototype did this very well  (it also allowed alternative
> > ways of entering data, e.g. by orgunit or period, another feature lacking
> > from DHIS  - though not a high priority, I think).
> > Knut
> >
> >>
> >> Yes, we have most of what is needed already, which hopefully means
> >> improving it would not be too much work.
> >> The reason the data entry in itself is not sufficient is that it only
> >> offers the one-dimensional list - for a data browser you would like to
> >> compare many orgunits for a period or many periods for an orgunit or
> many
> >> datasets for one period/orgunit combo - and this is indeed the kind of
> thing
> >> the current web pivot can do.
> >> So what would greatly enhance the usability of DHIS2 would be
> integration
> >> of the Data Entry and the Web Pivot (i.e. links to allow you to quickly
> get
> >> screens where you see your recently added data in context - and then we
> can
> >> add intuitive links from the Data Summary (currently misnamed Data
> browser)
> >> to this integrated Data Browser.
> >> So we are basically talking about intelilgent links between what is
> >> currently three separate modules. This would reduce clicking
> dramatically
> >> and make it easier to get an understanding of what is available.
> >> As a further step, I think it would be great if what the user has
> selected
> >> to see in the Web Pivot could then also be made into a report table with
> the
> >> click of a button. We have a lot of good parts that could become very
> >> powerful if linked better.
> >> Knut
> >>
> >>>
> >>> Me too thinks that extending the web pivot to use aggregated data
> values
> >>> and entered data values (with data set filter instead of group) is a
> good
> >>> idea that would not require too much development time. We could put
> this on
> >>> the list.
> >>>
> >>> Lars
> >>>
> >>>
> >>> _______________________________________________
> >>> Mailing list: https://launchpad.net/~dhis2-devs
> >>> Post to     : dhis2-devs@xxxxxxxxxxxxxxxxxxx
> >>> Unsubscribe : https://launchpad.net/~dhis2-devs
> >>> More help   : https://help.launchpad.net/ListHelp
> >>>
> >>
> >>
> >>
> >> --
> >> Cheers,
> >> Knut Staring
> >
> >
> >
> > --
> > Cheers,
> > Knut Staring
> >
> > _______________________________________________
> > Mailing list: https://launchpad.net/~dhis2-devs
> > Post to     : dhis2-devs@xxxxxxxxxxxxxxxxxxx
> > Unsubscribe : https://launchpad.net/~dhis2-devs
> > More help   : https://help.launchpad.net/ListHelp
> >
> >
>



-- 
Cheers,
Knut Staring