A pi filter is a circuit with two capacitors and an inductor. The inductor is inline in the supply voltage (between the input voltage and the input of the regulator. One capacitor is between ground and the input voltage, while the other one is between the input of the regulator and ground.
The first capacitor (between ground and input voltage) is for filtering noise going back to the supply. In our scenario, the noise is coming from the supply, so this capacitor is not needed.
We just removed one ``leg'' of the symbol pi, making it look like gamma.
The purpose of the inductor is to stop any sudden increase or decrease of current. Note that an inductor is more effective at this than a capacitor. The capacitor between the input of the regulator and ground acts as an additional bumper to soak up additional shock energy of the sudden change.
As an analogy, the inductor is the size and mass of a tire, whereas the capacitor is the spring of a suspension.
Of course, the next logical question is, ``how do we calculate the values of the capacitor and inductor?'' The answer to this question depends on many factors, including the average current consumed by the regulator (load current), the magnitude of the input voltage step, the tolerance of the regulator and etc.
Instead of purchasing different inductors for testing, you can use the equation at http://home.earthlink.net/~jimlux/hv/wheeler.htm or http://www.ee.surrey.ac.uk/Workshop/advice/coils/air_coils.html. Air core inductors are nice because they are easy to make from insulated solid-core copper wires. Choose the thickness that suits your current requirements.