Just as others have mentioned, the number inside your box is not an atomic but rank-1, so it adds an extra axis during unbox. ]a=. 0;0;1;1;1 +-+-+-+-+-+ |0|0|1|1|1| +-+-+-+-+-+ >a 0 0 1 1 1 ],&.>a +-+-+-+-+-+ |0|0|1|1|1| +-+-+-+-+-+ >,&.>a 0 0 1 1 1
You can use monadic ; which will unbox numbers as an rank-1 vector ; ,&.>a 0 0 1 1 1 It doesn't matter whether the boxed array has more columns or rows, J can handle boxed arrays of over tens of million of boxed cells. However if you can redesign your program, storing data as inverted tables, efficiency will be orders of magnitude better. That said, 50K boxed cells are very small, efficiency doesn't matter. On Tue, Jun 30, 2020 at 10:27 AM HH PackRat <[email protected]> wrote: > Hello, again! > > I feel a little stupid for asking for help with something probably so > elementary that I'm missing it for some reason. I've done lots of > switching between boxed and unboxed data in the past, but this one's > got me stumped. I would have attached the 28K data file, but > attachments aren't permitted. I considered tagging the data onto the > end of this message as more of the text of the message, but I figured > that wouldn't be allowed either. So anyone who would like to work > with the data themselves can contact me via email, and I will be happy > to send the data. (It's stock data from 2015 to 2020 with 2 columns: > date and an indication of up and down movement.) > > The raw data file looks like this: > > 2015-01-02,1 > 2015-01-03,1 > 2015-01-04,1 > 2015-01-05,0 > . . . > > Here's the J code I used: > > data=. [the data file read in from the computer] NB. 2 columns of boxed > data > > a=. |: data NB. transpose data to 2 rows (1982 columns, in this case) > > This is "a": > ┌───────┬───────┬───────┬───────┬─ > │2015-01-02│2015-01-03│2015-01-04│2015-01-05│ > ├───────┼───────┼───────┼───────┼─ . . . > │1 │1 │1 │0 │ > └───────┴───────┴───────┴───────┴─ > > a2=. 1{a NB. get data from row 1 > > This is "a2": > ┌─┬─┬─┬─┬ > │1 │ 1│1│ 0│ . . . > └─┴─┴─┴─┴ > > a3=. > a2 > > The following SHOULD be "a3" (as far as I know): > 1 1 1 0 . . . > > but "a3" ends up like this: > 1 > 1 > 1 > 0 > . . . > > And if I transpose this "a3", I end up with this: > 1110 . . . > > > Small "live" examples (not a script) work fine: > > ]a2=. 1;1;1;0;0;0 > > ┌─┬─┬─┬─┬─┬─┐ > │ 1│ 1│1│ 0│ 0│ 0│ > └─┴─┴─┴─┴─┴─┘ > > ]a3=. > a2 > > 1 1 1 0 0 0 > > > As I said, I'm stumped by this, and (for pattern matching purposes) > the remainder of my program depends on a row (not column) of 1's and > 0's (although, if necessary, I'm sure I could code the other way). > I'll probably be embarrassed by how simple the explanation is, but at > least I'll be able to continue work on the rest of my program. > > Harvey > > P.S. Does J have an upper limit on the number of items in rows? > Maybe that's the reason it (automatically?) transposed the data > vertically? This "small" file has 1982 items. The "full" file (once > I get things working correctly) has more than 14,700 items (58 years > of data) and other future files could easily go back 135 years (or > more than 49,300 items). Does J prefer lots of rows rather than lots > of columns? (I mean, we're talking about the possibility of 14,700+ > columns of data for this example--or possibly up to 49,300+ columns of > data in the future.) I was under the impression that J didn't care > either way, but maybe I'm wrong, and it does care? > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
