Re: st: Graphing long data with nested components

[Nick Cox <njcoxstata@gmail.com>]

No section of the manual is really geared to helping you identify what
graph to show if you don't know already.

For even 5 services, there are 2^5 - 1 possibilities for which of the
5 services are provided: the subtraction of 1 on the presumption that
no hospital provides no services has naturally a minute effect. For
147 services, the number is 2^147 - 1, of the kind usually called
astronomical, although perhaps it would better to say economic.

Of course, in practice, they won't all occur, but the explosion of
possibilities is likely to be a severe limit on what can be plotted.

I think you need to say more about what you want to do. Part of this
reads as if you perhaps intend some kind of cluster analysis, for
which see -cluster-. You could reduce your data to how many services
are provided, by each combination of hospital and PBC, but that
doesn't seem to be what you want either.

Nick

On Tue, Aug 14, 2012 at 1:06 PM, Anne McCarthy <AMcCarthy@hrb.ie> wrote:
> Dear statalisters,
>
> I have a question regarding graphing data in Stata/MP 10.1 for windows.
> The dataset details services provided by hospitals managed under public benefit corporations.
> Each public benefit corporation (PBC) may manage between 1 to 13 hospitals.
> A list of 147 services were identified across the entire range of hospitals and thus each row of the original dataset consisted of:
> a public benefit corporation, each hospital under the umbrella of this PBC, and the 147 services which could be potentially provided at each hospital.
> Initially I extracted the data into rows (as I did not know what the nature or distribution of hospitals or services was) I then reshaped the data, to give the structure below(a simple panel)- I have limited the example no. of hospitals to 5 and services to 4.
>
> Public benefit corporation   hospital   Service Accident and emergency     Service Alcohol addiction  Service  Allergy services  Service  Anaesthesia services
> a                               1               yes
> a                               2
> a                               3                                               yes
> a                               4
> a                               5
> b                               1
> b                               2                                               yes
> b                               3
> b                               4
> b                               5                                                                               yes                     yes
> c                               1                                               yes
> c                               2
> c                               3
> c                               4               yes
> c                               5
>
>
> I would like to be able to graphically illustrate the various cluster types observed in each PBC -nesting if possible service, within hospital, within PBC
> Any ideas on what I should look for in the manual to start this process
>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/