I did a test on a ship and put two GPS receivers horizontally. Afterwards I plotted the coordinates in a graph in Excel. The GPS receivers receive data asynchronously. If the interpolation is done correctly, the heading could be figured out by using the other sets of data the GPS receivers provided.

If a collection of posit data could be kept on the MCU using Regression will remove the outliers. Perhaps a Polynomial Regression would work a bit better, being better able to adjust to sudden and deep angular changes.

I used a ESP32 to store 96 hours worth of air pressure readings to predict the AIr Pressure trends for the next 5 days. Worked really well. I've since changed overt to using a RPi.

To accurately measure the distance between the GPS receivers. As it is now, the program calculates the GPS receivers diagonally, which make the distance bigger than it actually is.

I want to plot the route of the ship as neatly as possible. The distance is already known, but the system prints out something else, thanks for the link!

Well you could look at the link the several good regression libraries I posted in post number 4 to see how it works. You can look up the Python library scikit-learn to learn about LR. Many pieces of documentation about the Python scikit-learn library go into depth on how regression works.