Re: Smart vsum

[Juan Pechiar <juan@pechiar.com>]

[-- Attachment #1: Type: text/plain, Size: 1052 bytes --]

Hi Cecil.

The attached email from orgmode mailing list includes code for a
conditional vsum, which is what you need here.

Regards,
.j.

On Mon, Nov 15, 2010 at 01:10:52PM +0100, Cecil Westerhof wrote:
> It is just a table with activities and the needed time pro activity.
> When there is an asterisk in the first column, that means that the
> activity is already done. I would like a little more fancy table that
> also displays the time needed for the activities that still have to be
> done. Like the following:
> |---+----------+-----------|
> |   | activity | time      |
> |---+----------+-----------|
> | * | A        | 0@ 10'    |
> | * | B        | 0@ 05'    |
> |   | C        | 0@ 05'    |
> |   | D        | 0@ 05'    |
> |   | E        | 0@ 05'    |
> |---+----------+-----------|
> | # | Totaal   | 0.50      |
> | ^ |          | totalTime |
> | # | Totaal   | 0.25      |
> | ^ |          | restTime  |
> |---+----------+-----------|
> #+TBLFM: $totalTime = deg(vsum(@-II..@-I)); f2
>
> Is this possible? If yes, how would I do this?
>

[-- Attachment #2: Type: message/rfc822, Size: 7384 bytes --]

From: Christian Moe <mail@christianmoe.com>
To: Juan <Pechiar@computer.org>
Cc: emacs-orgmode@gnu.org, Inquisitive Scientist <inquisitive.scientist@gmail.com>
Subject: Re: [Orgmode] questions about table mode and spreadsheets
Date: Tue, 07 Sep 2010 11:20:59 +0200
Message-ID: <4C8603FB.9080409@christianmoe.com>

On 9/7/10 2:56 AM, Juan wrote:
> A very complex way of not adding the extra column:
>
> | name | a  | b | c |
> |------+----+---+---|
> | foo  | 1  | 2 | 3 |
> | bar  | 3  | 2 | 1 |
> | bar  | 4  | 5 | 6 |
> |------+----+---+---|
> |      | 7  |   |   |
> #+TBLFM: @5$2='(apply '+ (mapcar* (lambda(x y) (if (string= x "bar") y 0)) '(@I$1..@II$1) '(@I$2..@II$2)));L
>
>   * the two arguments at the end are the name and a columns: '(foo bar bar) and '(1 3 4)
>   * the lambda function returns the second argument if first is "bar", 0 otherwise.
>   * mapcar* applies the lambda function to arguments from the 2 lists.
>   * apply '+ adds the resulting list
>
> Regards,
> .j.

Neat! This is what I wanted to achieve. Good thing I gave up, though,
I see it would have kept me up all night.

(And yes, I meant "add a new /column/", not row.)

If one wants to do this often (e.g., in the other two columns), one
could tuck away some of the complexity into one's .emacs, and at the
same time get away from hard-coding the match string, like so:

#+begin_src elisp
   (defun vsumif (string x y)
   "Sum values of Y for all X matching STRING."
     (apply '+
            (mapcar*
             (lambda(x y)
               (if (string= x match) y 0))
             x y)))
#+end_src

Now, one can e.g. put the string one is matching for in the table. Try
updating the spreadsheet below, then changing `foo' in the bottom row
(@5$1) to `bar' and updating again.

| name | a | b | c |
|------+---+---+---|
| foo  | 1 | 2 | 3 |
| bar  | 3 | 2 | 1 |
| bar  | 4 | 5 | 6 |
|------+---+---+---|
| bar  |   |   |   |
#+TBLFM: @5$2='(vsumif '@5$1 '(@I$1..@II$1) '(@I$2..@II$2));L

It's still a lengthy formula and not the easiest to write. If you'd
like to add up all foos or bars for columns a, b and c, you may be
better off swapping rows and columns so you can use column formulas:

|   | name | foo | bar | bar | bar |
|---+------+-----+-----+-----+-----|
| / | <>   |   < |     |   > |  <> |
|   | a    |   1 |   3 |   4 |     |
|   | b    |   2 |   2 |   5 |     |
|   | c    |   3 |   1 |   6 |     |
#+TBLFM: $6='(vsumif '@1$6 '(@1$3..@1$5) '($3..$5));L

Again, replace `foo' in @1$6 with `bar' to get totals for bar.

I have added vertical lines to the table.

One could presumable write =vsumif= in a more general form that would
not be hard-coded to test only for matching strings. The function
defined above might be better named =vsumifstring=.

Cheers,
CM
>
> On Tue, Sep 07, 2010 at 12:44:03AM +0200, Christian Moe wrote:
>> On 9/6/10 3:38 PM, Inquisitive Scientist wrote:
>>
>>>    2. How do I compute the sum of a column only if a corresponding row
>>> matches some condition? For example, how do I compute the sum of
>>> numbers in column a for which the name in column "name" is "bar"? For
>>> example, I should get 7 for the sum in column a in the table below:
>>>
>>> | name | a | b | c |
>>> |------+---+---+---|
>>> | foo  | 1 | 2 | 3 |
>>> | bar  | 3 | 2 | 1 |
>>> | bar  | 4 | 5 | 6 |
>>> |------+---+---+---|
>>
>> Here's one way: Add a new row

sorry: meant "add a new column"

>> after the first, as below. Then run C-c
>> C-c on the formula line:
>>
>> | name |   | a | b | c |
>> |------+---+---+---+---|
>> | foo  |   | 1 | 2 | 3 |
>> | bar  |   | 3 | 2 | 1 |
>> | bar  |   | 4 | 5 | 6 |
>> |------+---+---+---+---|
>> |      |   |   |   |   |
>>    #+TBLFM: $2='(if (string= $1 "bar") 1 0)::
>> @5$3=vsum(vmask(@I$2..@II$2,@I..@II))
>>
>> It does exactly what you asked, but I don't think it will scale well...
>>
>
> _______________________________________________
> Emacs-orgmode mailing list
> Please use `Reply All' to send replies to the list.
> Emacs-orgmode@gnu.org
> http://lists.gnu.org/mailman/listinfo/emacs-orgmode
>


[-- Attachment #3: Type: text/plain, Size: 201 bytes --]

_______________________________________________
Emacs-orgmode mailing list
Please use `Reply All' to send replies to the list.
Emacs-orgmode@gnu.org
http://lists.gnu.org/mailman/listinfo/emacs-orgmode