A VNA does that alone. The network can be reconstructed from reflection/transmission properties and assuming the network is linear and passive (which should be a fine assumption here).
Curiously, this works even if the network is not made of RLC lumped components; in that case, you get an approximation (e.g., a transmission line as a huge LC ladder).
Real components aren't ideal RLC lumps either, so you can learn quite a lot about the components in the tuner for the same reason (e.g., not just the capacitor's capacitance, but ESR, ESL and higher order aspects as well).
You can replace the VNA with a signal source, reflectance bridge, and various loads; and a lot more fiddling.
In either case, solving the network requires simultaneous equations in N variables for N elements, and also something about least squares regression for overdetermined, noisy systems (which is to say, the VNA grabs a huge swath of data, way more points than strictly required to solve the problem, and few of which actually match up exactly with any particular solution -- that is, they're somewhat noisy). So, you're kind of SOL on the math bit, to go any farther.
Tim