There is probably no published figure on the accuracy of DNA testing because it varies every time.
Researchers of course don't look at the whole strand of DNA, which is trillions of base pairs long. They look at slices, fragments, tiny runs of a couple hundred pairs only. Such is the nature of DNA evidence collected at crime scenes.
Using modern high speed computers they can sequence these fragments, and compare them to known sequences in the publicly-published DNA catalogs. They can tell if the DNA is human, or canine, or beetle by the number of congruences with known DNA profiles.
The amount of usable DNA varies considerably, as does the success in finding matches.
No two DNA searches come up with the same result.
Only if a person has a known DNA anomaly, and only if DNA is collected that just happens to contain the fragment of DNA with that anomaly, can DNA testing be used to positively identify an individual. Such an occurrence is so rare as to be vanishing, percentage wise.
Remember, humans share 96% of our DNA with chimpanzees. Within our species, we share something like 99.86%. Finding identifying markers is very very hard.