The sheet calculates the end reactions Fa and Fb, whether correctly or not I haven't tried to determine.
His solution method is to solve the differential equation for deflection from both sides, which is valid (and impressive.) He then creates a deflection function by patching these two solutions together and takes the double derivative to get moment; again correctly. To get shear he takes the derivative of moment, which works everywhere except at the ends of the beam and where the two solutions patch together, where the derivative fluctuates wildly.
Mathcad scales the shear graph to show all of it's reported values, so it shows the wildly incorrect fluctuations. If we rescale the shear graph to reasonable limits and ignore the wild excursions it should make sense.