GPS Data are already heavily Kalman filtered. This is done inside the GPS receiver.
Dont expect an accuracy gain in position (lat / lon) if you create your own kalman filter.
Further you dont have the information that the internal GPS receiver had. It feeds its internal kalman filter 1000 time / per second, before it outputs one location.
In your own post processing filter you might gain a smoother track (related to visualizing the positions). But smoother is not more accurate, only more pleasant to view.
Another topic is whether or not the GPS positions must be available in real time, e.g display the current position on screen. If you want to smooth your tracks afterwards (non realtime) you could be successfull, but I would not use a kalman filter for that case.
A kalman filter is well suited for real time filtering, for post processing, you could try a sliding average with a triangle window filter (easy to implement, while kalman is very komplex)
