That wouldn't make a lot of sense, though, for classic operations on
sparse matrices (like +/ .*), would it?
That said, it might sense to leave 0 as the sparse element, and then
use the distinction between sparse zeros and dense zeros to track
undefined values...
Thanks,
--
Raul
On Wed, Sep 2
I thought that possibly sparse matrices might be a way to handle missing
values, particularly if there a lot of undefined values. But couldn't
figure out how to specify the sparse element to be _. .
On Wed, Sep 20, 2017 at 3:22 PM, Henry Rich wrote:
> No; see
>
> http://code.jsoftware.com/wiki/V
No; see
http://code.jsoftware.com/wiki/Vocabulary/under Details
http://code.jsoftware.com/wiki/Vocabulary/percent#dyadic Details
It would perhaps be better if 0 * _5 gave -0, but it doesn't. The other
deviations are improvements added by J.
Henry Rich
On Wed, Sep 20, 2017 at 10:51 AM, Erli
My memory of grappling with the "how to handle missing data" issue in APL
and J was that the most complete solution was to have a separate boolean
array of the same shape as the data array that identified missings. This is
certainly a clunkier approach from the user's POV compared to R, but is
argu
I'm probably missing the point, but it's perhaps worth observing that
you might choose to consider ALL the numbers here (except the exponent)
as belonging to the finite field of numbers mod 13, in which case
"Roger's approach" doesn't fail.
What about 10^20 ?
13&|@(10&^) 20
9
(in fact,
13&|
I do not have any problem with using _ or _. or __ as placeholders
where you do not know what value to use.
I do not think, however, that distinguishing between different kinds
of _. values is something we should have to deal with.
Thanks,
--
Raul
On Wed, Sep 20, 2017 at 11:04 AM, Vijay Lulla
I've used _ and __ as placeholders for missing or invalid data. I then
filter them out as needed in calculations. It's simple and has worked out
fine for me.
On Wed, Sep 20, 2017 at 11:22 AM, Erling Hellenäs
wrote:
> Hi all!
>
> Programs which acted on arrays containing missing data would need
>
Hi all!
Programs which acted on arrays containing missing data would need
definitions which accounted for this missing data. Take a look at R. I
think most languages have ways of handling missing data, even if
calculations do not always work directly on arrays containing missing
data, like t
To answer Henry's question based on the behavior of R, which Erling
advocates, and I too love!!
In R, length of vector (what we'd call list in J) with missing values is
still the actual number of elements in a vector. Missing value *is* still
a value in a list and it gets counted! However, it ma
Hi all !
Here is part of the standard. Required exception handling. Things I
discuss in my post. Is it implemented in J?
https://en.wikipedia.org/wiki/IEEE_754#Exception_handling
The floating point standard is obviously used in environments where the
users can not afford random results, so the
If a list (a) can contain 'missing' entries, what does (# a) mean? What
happens to
mean =: +/ % #
?
How is u/\. a defined?
Henry Rich
On Wed, Sep 20, 2017 at 10:25 AM, Erling Hellenäs
wrote:
> Because you possibly have an array where some cells have missing data and
> you want to run your f
I hope others want to read my post even if Raul discards it as not
worthy of comments. /Erling
Den 2017-09-20 kl. 12:25, skrev Raul Miller:
This isn't just J - this is the IEEE-754 floating point standard.
It would be really nice if computers could deal with infinities,
instantly, at no cost.
Because you possibly have an array where some cells have missing data
and you want to run your functions directly against this array without
caring about the possibly missing data. Thats how I think R works. All
functions work with cells where the data is missing. Correct me if I'm
wrong. R mig
Er... ok: yes, not exactly but sort of, and that depends?
It's a messy subject, but switching to extended precision is a much
more general approach for this kind of problem.
Thanks,
--
Raul
On Wed, Sep 20, 2017 at 5:31 AM, Erling Hellenäs
wrote:
> Hi all !
>
> It just surprised me that you di
Why not just use another number? Many more bits that way...
Thanks,
--
Raul
On Wed, Sep 20, 2017 at 5:47 AM, Erling Hellenäs
wrote:
> Hi all !
>
> "A NaN may carry a payload that is intended for diagnostic information
> indicating the source of the NaN"
>
> https://en.wikipedia.org/wiki/IEEE_
This isn't just J - this is the IEEE-754 floating point standard.
It would be really nice if computers could deal with infinities,
instantly, at no cost. Sadly, though, that's not going to happen.
FYI,
--
Raul
On Wed, Sep 20, 2017 at 5:21 AM, Erling Hellenäs
wrote:
> Hi all!
>
> This is how
Hi all !
"A NaN may carry a payload that is intended for diagnostic information
indicating the source of the NaN"
https://en.wikipedia.org/wiki/IEEE_754#Alternate_exception_handling
Maybe this is a way to represent missing data?
Cheers,
Erling
Den 2017-09-20 kl. 11:21, skrev Erling Hellen
Hi all !
It just surprised me that you didn't mention it as a possible solution
to the problems of Skip Cave. Since you did not it seemed there might be
a problem with this solution. It might not be a possible solution.
That's why I ask these three questions. It's not because I can not read
o
Hi all!
This is how J handles overflow and underflow. This is exacly how the
platform handles it. JWithATwist gives exactly the same results. As we
can see overflow results in infinity, underflow results in a zero result.
0.1234567891234567e_15 * 0.1234567891234567e_307
0
0.123456789123
19 matches
Mail list logo
