Re: Text output and WIDTH

[Ben Pfaff]

There is supposed to be some logic to it. Minimum and maximum are always
values from the data, so it makes sense to display them in the data's own
format.  I would have guessed that we do the same for mean, but not
necessarily for standard deviation. There may be room for improvement here;
I went through and assigned most of these a few months ago (based on what
was there already but I did make some new judgments) and I might have
gotten some of them wrong.

On Thu, Jan 9, 2020 at 12:20 AM Frans Houweling <address@hidden> wrote:

> I see the point. The variables are indeed F8.2 format. DESCRIPTIVES
> created these from F12.4 input variables.
> Problem solved. Thanks.
> It is interesting that, as you can see from the second DESCRIPTIVES,
> Mean and Std Dev do seem to honor SET FORMAT, while Minimum and Maximum
> do not.
> frans
>
>
> DESCRIPTIVES meanV TO minIR /SAVE.
>
> Mapping of Variables to Z-scores
> +--------------+---------------+
> |    Source    |     Target    |
> +--------------+---------------+
> |meanV         |ZmeanV         |
> |maxV          |ZmaxV          |
> |minV          |ZminV          |
> |meanDistribV  |ZmeanDistribV  |
> |contrastV     |ZcontrastV     |
> |entropyV      |ZentropyV      |
> |secondAngularV|ZsecondAngularV|
> |meanIR        |ZmeanIR        |
> |maxIR         |ZmaxIR         |
> |minIR         |ZminIR         |
> +--------------+---------------+
>
>                       Descriptive Statistics
> +--------------------+----+--------+--------+--------+---------+
> |                    |  N |  Mean  | Std Dev| Minimum| Maximum |
> +--------------------+----+--------+--------+--------+---------+
> |meanV               |1024|  6.3672|  7.1813|  1.0000|  96.0000|
> |maxV                |1024|127.5439| 60.0359|  2.0000| 255.0000|
> |minV                |1024| 40.5364| 30.8521|  1.0273| 167.1484|
> |meanDistribV        |1024|   .0568|   .0350|   .0000|    .1747|
> |contrastV           |1024|538.8151|474.4201|   .0000|3211.4753|
> |entropyV            |1024|   .1233|   .1968|   .0145|   1.0000|
> |secondAngularV      |1024|  3.1392|  1.0143|   .0000|   4.3828|
> |meanIR              |1024|186.0801| 33.3433|  1.0000| 250.0000|
> |maxIR               |1024|243.1426|  8.5934|148.0000| 255.0000|
> |minIR               |1024|223.3901| 21.6168| 80.3164| 252.9453|
> |Valid N (listwise)  |1024|        |        |        |         |
> |Missing N (listwise)|   0|        |        |        |         |
> +--------------------+----+--------+--------+--------+---------+
>
> DESCRIPTIVES ZmeanV TO ZminIR.
>
>                      Descriptive Statistics
> +-------------------------+----+-----+-------+-------+-------+
> |                         |  N | Mean|Std Dev|Minimum|Maximum|
> +-------------------------+----+-----+-------+-------+-------+
> |Z-score of meanV         |1024|.0000| 1.0000|   -.75|  12.48|
> |Z-score of maxV          |1024|.0000| 1.0000|  -2.09|   2.12|
> |Z-score of minV          |1024|.0000| 1.0000|  -1.28|   4.10|
> |Z-score of meanDistribV  |1024|.0000| 1.0000|  -1.62|   3.37|
> |Z-score of contrastV     |1024|.0000| 1.0000|  -1.14|   5.63|
> |Z-score of entropyV      |1024|.0000| 1.0000|   -.55|   4.45|
> |Z-score of secondAngularV|1024|.0000| 1.0000|  -3.09|   1.23|
> |Z-score of meanIR        |1024|.0000| 1.0000|  -5.55|   1.92|
> |Z-score of maxIR         |1024|.0000| 1.0000| -11.07|   1.38|
> |Z-score of minIR         |1024|.0000| 1.0000|  -6.62|   1.37|
> |Valid N (listwise)       |1024|     |       |       |       |
> |Missing N (listwise)     |   0|     |       |       |       |
> +-------------------------+----+-----+-------+-------+-------+
>
>
>
>
> On 1/9/20 8:54 AM, Ben Pfaff wrote:
> > You can get more digits by changing the formats of the variables
> > involved, which I guess currently have F8.2 format or some similar
> > format with 2 decimal places.
> >
> > On Wed, Jan 8, 2020 at 11:46 PM Frans Houweling <address@hidden>
> wrote:
> >> Hi Ben, it's the "Final Cluster Centers" table I am interested in.
> >>
> >>
> >>                  Final Cluster Centers
> >> +-------------------------+----------------------+
> >> |                         |        Cluster       |
> >> |                         +-----+----+-----+-----+
> >> |                         |  1  |  2 |  3  |  4  |
> >> +-------------------------+-----+----+-----+-----+
> >> |Z-score of meanV         | -.43| .06| 3.57| 8.58|
> >> |Z-score of maxV          |-1.04| .59| 1.10| 1.23|
> >> |Z-score of minV          | -.94| .46| 2.41| 3.28|
> >> |Z-score of meanDistribV  |-1.06| .62|  .71| -.15|
> >> |Z-score of contrastV     | -.90| .54|  .44| -.51|
> >> |Z-score of entropyV      |  .77|-.45| -.49| -.43|
> >> |Z-score of secondAngularV|-1.01| .60|  .65|  .28|
> >> |Z-score of meanIR        |  .96|-.52|-1.28|-2.76|
> >> |Z-score of maxIR         |  .57|-.19|-2.48|-7.91|
> >> |Z-score of minIR         |  .89|-.43|-2.19|-4.20|
> >> +-------------------------+-----+----+-----+-----+
> >> But it's not that important - I was just wondering why I didn't get more
> >> digits.
> >> Thanks
> >>
> >> frans
> >>
> >>
> >> On 1/9/20 8:33 AM, Ben Pfaff wrote:
> >>> At a glance, it appears that QUICK CLUSTER output contains only two
> >>> kinds of values:
> >>> - Values that are always integers. PSPP doesn't honor SET FORMAT for
> >>> these kinds of values since the decimal places would not be useful.
> >>> - Values of particular variables. PSPP uses the variables' own print
> >>> formats for these values.
> >>> Which values' formats did you want to control?
> >>>
> >>> On Wed, Jan 8, 2020 at 12:52 PM Frans Houweling <address@hidden>
> wrote:
> >>>> While you're at it, you might check why SET FORMAT has no effect on
> the
> >>>> output of QUICK CLUSTER (it has on LOGISTIC REGRESSION). I hoped -O
> >>>> format=F9.4 would work, but it doesn't.
> >>>> Thanks
> >>>> frans
> >>>>
> >>>>
> >>>> On 1/8/20 8:21 AM, John Darrington wrote:
> >>>>> I thought that SET WIDTH should also work.
> >>>>>
> >>>>> Also I thought that when run interactively from a terminal, it should
> >>>>> default to the width of the terminal.
> >>>>>
> >>>>> Perhaps we need to investigate why this is not happening.
> >>>>>
> >>>>> J'
> >>>>>
> >>>>>
> >>>>> On Mon, Jan 06, 2020 at 09:58:59PM -0800, Ben Pfaff wrote:
> >>>>>         On Mon, Jan 6, 2020 at 12:54 PM Frans Houweling <
> address@hidden> wrote:
> >>>>>         >    is there any way to prevent text output to wrap at 80
> columns?
> >>>>>         > I would like to retrieve cluster centers from the output
> of QUICK
> >>>>>         > CLUSTER, but this is hindered by wrapping:
> >>>>>
> >>>>>         Yes, you can specify the maximum output width with "-O
> width=200" on
> >>>>>         the command line for 200 columns (or whatever width you
> like).
>