[Jprogramming] 2-dimensional summary report

[Roger Hui]

It's more efficient to work with the indices rather
than the labels.  You can put the labels back in
at the end for the report.



----- Original Message -----
From: June Kim <juneaftn at gmail.com>
Date: Saturday, May 17, 2008 11:16
Subject: Re: [Jprogramming] 2-dimensional summary report
To: Programming forum <programming at jsoftware.com>

> 2008/5/18, Roger Hui <rhui000 at shaw.ca>:
> > Convert areas and models into indices and assign sales
> > into those locations of a matrix of zeros.
> >
> > ij=. area,.model
> > (ij +//. sales) (<"1 ~.ij)}(m,n)$0
> 
> Thanks.
> 
> It is basically similar to what I originally wrote, but yours is 
> much cleaner.
> 
> Hence, I have cleaned up my code as following:
> 
>    load 'strings'
>    data=: 0 : 0
> area   model sales
> north  T2    3
> south  T2    4
> north  T3    10
> north  T3    5
> east   T1    1
> west   T3    1
> south  T2    11
> north  T4    2
> west   T3    1
> east   T4    1
> west   T4    6
> )
>    d=: (<a:;<2) ((".&.>)@{)`[`]} }. cut;._2 data
> 
>    ind=:i.~ ~.
>    areas=: ind 0{"1 d
>    models=:ind 1{"1 d
>    sales=: > 2{"1 d
> 
>    keys=:areas,.models
>    salessum=:keys +//. sales
> 
>    blank=:0$~(#~.areas),#~.models
>    salessum (<"1 ~. keys)} blank
>  3 15 0 2
> 15  0 0 0
>  0  0 1 1
>  0  2 0 6
> 
> 
> Any more suggestions for improvement?
> 
> >
> > If there can be at most one figure for each
> > pair of (area,model) then you can skip the
> > cumulation and the nub:
> >
> > sales (<"1 area,.model)}(m,n)$0
> >
> 
> Right, but it's not appicable to my case.
> 
> >
> >
> > ----- Original Message -----
> > From: June Kim <juneaftn at gmail.com>
> > Date: Saturday, May 17, 2008 7:44
> > Subject: [Jprogramming] 2-dimensional summary report
> > To: Programming forum <programming at jsoftware.com>
> >
> > > Suppose the data is given as following:
> > >
> > >    data=: 0 : 0
> > > area   model sales
> > > north  T2    3
> > > south  T2    4
> > > north  T3    10
> > > north  T3    5
> > > east   T1    1
> > > west   T3    1
> > > south  T2    11
> > > north  T4    2
> > > west   T3    1
> > > east   T4    1
> > > west   T4    6
> > > )
> > >    d=: (<a:;<2) ((".&.>)@{)`[`]} }. 
> cut;._2 data
> > >
> > >
> > > What I want is a report of sales for area by model:
> > >
> > >
> > > NB.          
> T2   T3   T1  T4
> > >    NB. north     3
> > > 15    0   2
> > >    NB. south    15
> > > 0    0   0
> > >    NB. east
> > > 0    0    1   1
> > >    NB. west
> > > 0    2    0   6
> > >
> > >
> > > The final matrix doesn't need to contain row, column-headers 
> --
> > > it is
> > > fine that it contains only the sales total for each area and model
> > > pairs.
> > >
> > > I'd like to know easy and simple ways to getting that matrix 
> in J.
> > > What's your suggestion?