Table calculation

[Xiao-Yong Jin <xj2106@columbia.edu>]

Hi, in order to calculate the average with some invalid column, I
managed to write a short elisp script.  You can see the difference
between the `0' and the `-' in this table:

| ! |  P1 |  P2 | P3  |  P4 |  P5 | Average |
| # | 3.6 | 4.0 | 4.8 |   0 | 4.7 |    3.42 |
| # | 5.0 | 3.3 | -   | 4.7 | 4.0 |    4.25 |
#+TBLFM: $7='(let ((data '("$P1" "$P2" "$P3" "$P4" "$P5"))
                   (n 0)
                   (total 0))
                  (while data
                    (unless (string= (car data) "-")
                      (setq total
                        (+ total (string-to-number (car data))))
                      (setq n (1+ n)))
                    (setq data (cdr data)))
                  (/ total n));%.2f

It is reformatted to be more readable, since one cannot put a
multi-line expression in #+TBLFM.

I admit it is really painful to do this.  I want to know if anyone
knows how I can simplify this a bit, or if there is a short cut to do
this.  I'm not familiar with the internal implementation of org table,
but is there any possibility to make it as a built-in function that I
can put it somewhere in the same org file so that I can easily do some
thing like mymean($P1..$P5)?

Thanks,
Xiao-Yong
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---