Hi there,
Bug #24064 (submitted by [EMAIL PROTECTED]) requests a standard deviation
function for PHP. I realise that any of you could implement this in 10
minutes, but according to the bug database it is still Open so I figured I
would give it a try myself!
There are probably a dozen errors in the code and/or places where it could
be better optimised, but I'm hoping one of you might be able to help with
that. So, the attached diff file implements the function array_std_dev(),
to calculate standard deviation using the deviation method.
With the function in place, standard deviation is calculated like this:
<?php
$scores = array(18,5,7,18,3,2,10);
print array_std_dev($score);
// prints 6.6833125519211
?>
My first attempt at implementing this was using an extra array to buffer the
deviations - this was more out of curiosity to see how the array stuff
works. Sadly, it caused PHP to segfault and I couldn't figure out why - can
any of you help me spot the brain fart? (I've attached the offending code
in bad_stddev_code.txt)
Yours,
Paul
PS: I'm not on the internals list, so I would appreciate it if you would CC
me on your reply.
/* {{{ proto mixed array_std_dev(array input)
Returns the standard deviation of the array entries */
PHP_FUNCTION(array_std_dev)
{
zval **input,
**entry,
*entry_n,
*squareddev;
int argc = ZEND_NUM_ARGS();
HashPosition pos;
double dval;
double total;
double mean;
int numelements = 0; // note this is calcuated by hand, not using
zend_hash_num_elements()
double deviation;
if (argc != 1 || zend_get_parameters_ex(argc, &input) == FAILURE) {
WRONG_PARAM_COUNT;
}
if (Z_TYPE_PP(input) != IS_ARRAY) {
php_error_docref(NULL TSRMLS_CC, E_WARNING, "The argument should be an
array");
return;
}
ZVAL_LONG(return_value, 0);
total = 0;
// step one: sum the values of the array
for (zend_hash_internal_pointer_reset_ex(Z_ARRVAL_PP(input), &pos);
zend_hash_get_current_data_ex(Z_ARRVAL_PP(input), (void **)&entry,
&pos) == SUCCESS;
zend_hash_move_forward_ex(Z_ARRVAL_PP(input), &pos)) {
if (Z_TYPE_PP(entry) == IS_ARRAY || Z_TYPE_PP(entry) == IS_OBJECT)
continue;
entry_n = *entry;
zval_copy_ctor(entry_n);
convert_scalar_to_number(entry_n TSRMLS_CC);
convert_to_double(entry_n);
total += Z_DVAL_P(entry_n);
// this is incremented by hand so that it doesn't count object and
array elements as an element
numelements++;
}
// step two: calculate the mean of the input array
mean = total / numelements;
// step three: get the deviation from the mean of each number in the input
array, and add it to the squareddev array
MAKE_STD_ZVAL(squareddev);
array_init(squareddev);
for (zend_hash_internal_pointer_reset_ex(Z_ARRVAL_PP(input), &pos);
zend_hash_get_current_data_ex(Z_ARRVAL_PP(input), (void **)&entry,
&pos) == SUCCESS;
zend_hash_move_forward_ex(Z_ARRVAL_PP(input), &pos)) {
if (Z_TYPE_PP(entry) == IS_ARRAY || Z_TYPE_PP(entry) == IS_OBJECT)
continue;
entry_n = *entry;
zval_copy_ctor(entry_n);
convert_scalar_to_number(entry_n TSRMLS_CC);
convert_to_double(entry_n);
Z_DVAL_P(entry_n) -= mean;
zend_hash_next_index_insert(Z_ARRVAL_P(squareddev), entry_n,
sizeof(zval *), NULL);
}
// step four: sum the squared deviation array
total = 0;
for (zend_hash_internal_pointer_reset_ex(Z_ARRVAL_P(squareddev), &pos);
zend_hash_get_current_data_ex(Z_ARRVAL_P(squareddev), (void
**)&entry, &pos) == SUCCESS;
zend_hash_move_forward_ex(Z_ARRVAL_P(squareddev), &pos)) {
entry_n = *entry;
zval_copy_ctor(entry_n);
convert_scalar_to_number(entry_n TSRMLS_CC);
convert_to_double(entry_n);
total += Z_DVAL_P(entry_n);
}
// step five: divide the sum of the squared deviation array by the number of
elements - 1
total /= numelements - 1;
Z_DVAL_P(return_value) = sqrt(total);
}
/* }}} */
--
PHP Internals - PHP Runtime Development Mailing List
To unsubscribe, visit: http://www.php.net/unsub.php
