Hello Mark!
The 'Transformation matrix' parameter is a Float4 value. So the function graph applied to it needs to output that value type.
A simple way to connect an exposed Float parameter to the matrix's Y2 component is to build a Float4 value using a Vector Float4 node, which combines any pair of values that together add up to 4 Float components – E.g., Float and Float 3, Float2 and Float2. The final value is the result of appending the 'Last' input to the 'In' input.
Here is an example of such a setup:

The exposed Float parameter.

The function graph applied to the 'Transformation matrix' parameter.
Let me know if you have any questions about this process.
Best regards.