APL - quickest way?

Imagine a nested array like this:


ID     IND1    V1    IND2    V2    IND3    V3    DATE

1       A       100    -     -        -     -     2001-12-01
2       A        80    -     -        C    20     2001-12-01
3       B        70    C     5        D    10     2001-12-01

where - indicates a missing or unassigned value.

1. Quickest APL way to reformat it as follows?
(replacing missing values by zero)

ID     IND      V       DATE

1       A       100     2001-12-01
1       B         0     2001-12-01
1       C         0     2001-12-01
1       D         0     2001-12-01
2       A        80     2001-12-01
2       B         0     2001-12-01
2       C        20     2001-12-01
2       D         0     2001-12-01
3       A         0     2001-12-01
3       B        70     2001-12-01
3       C         5     2001-12-01
3       D        10     2001-12-01

2. Quickest APL way to reformat it as follows?
(lose rows with missing values)

ID     IND      V       DATE

1       A       100     2001-12-01
2       A        80     2001-12-01
2       C        20     2001-12-01
3       B        70     2001-12-01
3       C         5     2001-12-01
3       D        10     2001-12-01

On Mar 13, 10:44=A0am, AAsk <AA2e...@lycos.co.uk> wrote:
> Imagine a nested array like this:
>
> ID =A0 =A0 IND1 =A0 =A0V1 =A0 =A0IND2 =A0 =A0V2 =A0 =A0IND3 =A0 =A0V3 =A0 =[/color
]
=A0DATE
>
> 1 =A0 =A0 =A0 A =A0 =A0 =A0 100 =A0 =A0- =A0 =A0 - =A0 =A0 =A0 =A0- =A0 =
=A0 - =A0 =A0 2001-12-01
> 2 =A0 =A0 =A0 A =A0 =A0 =A0 =A080 =A0 =A0- =A0 =A0 - =A0 =A0 =A0 =A0C =A0 =[/color
]
=A020 =A0 =A0 2001-12-01
> 3 =A0 =A0 =A0 B =A0 =A0 =A0 =A070 =A0 =A0C =A0 =A0 5 =A0 =A0 =A0 =A0D =A0 =[/color
]
=A010 =A0 =A0 2001-12-01
>
> where - indicates a missing or unassigned value.
>
> 1. Quickest APL way to reformat it as follows?
> =A0 =A0(replacing missing values by zero)
>
> ID =A0 =A0 IND =A0 =A0 =A0V =A0 =A0 =A0 DATE
>
> 1 =A0 =A0 =A0 A =A0 =A0 =A0 100 =A0 =A0 2001-12-01
> 1 =A0 =A0 =A0 B =A0 =A0 =A0 =A0 0 =A0 =A0 2001-12-01
> 1 =A0 =A0 =A0 C =A0 =A0 =A0 =A0 0 =A0 =A0 2001-12-01
> 1 =A0 =A0 =A0 D =A0 =A0 =A0 =A0 0 =A0 =A0 2001-12-01
> 2 =A0 =A0 =A0 A =A0 =A0 =A0 =A080 =A0 =A0 2001-12-01
> 2 =A0 =A0 =A0 B =A0 =A0 =A0 =A0 0 =A0 =A0 2001-12-01
> 2 =A0 =A0 =A0 C =A0 =A0 =A0 =A020 =A0 =A0 2001-12-01
> 2 =A0 =A0 =A0 D =A0 =A0 =A0 =A0 0 =A0 =A0 2001-12-01
> 3 =A0 =A0 =A0 A =A0 =A0 =A0 =A0 0 =A0 =A0 2001-12-01
> 3 =A0 =A0 =A0 B =A0 =A0 =A0 =A070 =A0 =A0 2001-12-01
> 3 =A0 =A0 =A0 C =A0 =A0 =A0 =A0 5 =A0 =A0 2001-12-01
> 3 =A0 =A0 =A0 D =A0 =A0 =A0 =A010 =A0 =A0 2001-12-01
>
> 2. Quickest APL way to reformat it as follows?
> =A0 =A0(lose rows with missing values)
>
> ID =A0 =A0 IND =A0 =A0 =A0V =A0 =A0 =A0 DATE
>
> 1 =A0 =A0 =A0 A =A0 =A0 =A0 100 =A0 =A0 2001-12-01
> 2 =A0 =A0 =A0 A =A0 =A0 =A0 =A080 =A0 =A0 2001-12-01
> 2 =A0 =A0 =A0 C =A0 =A0 =A0 =A020 =A0 =A0 2001-12-01
> 3 =A0 =A0 =A0 B =A0 =A0 =A0 =A070 =A0 =A0 2001-12-01
> 3 =A0 =A0 =A0 C =A0 =A0 =A0 =A0 5 =A0 =A0 2001-12-01
> 3 =A0 =A0 =A0 D =A0 =A0 =A0 =A010 =A0 =A0 2001-12-01

If you want to do it in J you could do it like this

require'misc'
a=3D: 0 : 0
ID	IND1	V1	IND2	V2	IND3	V3	DATE
1	A	100	-	-	-	-	2001-12-01
2	A	80	-	-	C	20	2001-12-01
3	B	70	C	5	D	10	2001-12-01
)

b=3D:chop a
c=3D:TAB chop each b
d=3D:>{.c
e=3D:>}.c
(d)=3D:|:e

m1=3D:ID,.' ',.IND1,.' ',.V1,.' ',.DATE
m2=3D:ID,.' ',.IND2,.' ',.V2,.' ',.DATE
m3=3D:ID,.' ',.IND3,.' ',.V3,.' ',.DATE
m1,m2,m3

APL the quickest way?

i.e. APL not J

On Mar 13, 10:44 am, AAsk <AA2e...@lycos.co.uk> wrote:
> Imagine a nested array like this:
>
> ID     IND1    V1    IND2    V2    IND3    V3    DATE
>
> 1       A       100    -     -        -     -     2001-12-01
> 2       A        80    -     -        C    20     2001-12-01
> 3       B        70    C     5        D    10     2001-12-01
>
> where - indicates a missing or unassigned value.
>
> 1. Quickest APL way to reformat it as follows?
>    (replacing missing values by zero)
>
> ID     IND      V       DATE
>
> 1       A       100     2001-12-01
> 1       B         0     2001-12-01
> 1       C         0     2001-12-01
> 1       D         0     2001-12-01
> 2       A        80     2001-12-01
> 2       B         0     2001-12-01
> 2       C        20     2001-12-01
> 2       D         0     2001-12-01
> 3       A         0     2001-12-01
> 3       B        70     2001-12-01
> 3       C         5     2001-12-01
> 3       D        10     2001-12-01
>
> 2. Quickest APL way to reformat it as follows?
>    (lose rows with missing values)
>
> ID     IND      V       DATE
>
> 1       A       100     2001-12-01
> 2       A        80     2001-12-01
> 2       C        20     2001-12-01
> 3       B        70     2001-12-01
> 3       C         5     2001-12-01
> 3       D        10     2001-12-01

If you already have an apl array then those "missing" values are
already filled with something!
In Dyalog if the array is the result of some external process then
"null" values are likely to have been prefilled with the system
variable =E2=8E=95NULL. As I recall getting stuff in from SAS a couple of
decades ago empty cells were just =E2=8A=82''. Either way you need to know.

So let's say you have a function (Nulls) that, given the array,
returns a conforming boolean array that tells you which they are.

I could now present fifty trivial ways to do the actual merge (option
1) or compression (option 2) but which of them is quickest and in fact
which of them is possible depends on which APL and is almost certainly
going to be quicker than running our (Nulls) function anyway.