There are many reasons why this is wrong. The first one is the assumption of 1 supernova per 25 years. That was the value arrived at in Gustav Tammann's article published in 1970. Others got values up to 100 years (Tammann himself changed his value later). All these values are really only good guesses.
Worse errors are made in the percentage of remnants that should be visible. To start with, only a small percentage of them are visible; the others are obscured by dust in the galaxy. This does not happen to the same extent to the supernovas themselves, as they are intrinsically very much brighter.
Sarfati also uses an old paper to come up with the number of remnants in the Large Magellanic Cloud (a satellite galaxy to the Milky Way). However, that figure was considered to be wrong (and much smaller) in later studies. Older remnants are very difficult to distinguish against a background of dust in the galaxy. Many of them disappear completely after about 10,000 years or so.
It should be noted that, using Sarfati's own reasoning, we should not be able to see any remnants older than 6000 years. Instead, we know of remnants (e.g. G166.2 + 2.5) that are over 100,000 years old. How did those originate in his universe? The same goes for the distance to supernovas. The distance to SN 1987A has been trigonometrically measured at 167,000 light years (the paper says 51,200 parsecs). In other words, its light took more than 30 times longer to reach us than the creationist universe has supposedly existed.
Remember also that a supernova typically becomes either a pulsar or a black hole, both of which are very hard to observe. A black hole can only be "seen" through its effects on other matter, and a pulsar is mainly visible if the solar system happens to be in line with its sweeping radio beam (with some pulsars the star itself has also been seen). Hence, all we can see of them long after the event is an expanding cloud of gas that gets dimmer over the millennia.
There are several other reasons which are covered in a very good article on TalkOrigins. That article has links to all the original scientific papers referred to, so you can check the validity yourself.