Just keep in mind that Pickle is not human readable. If you want to check what's inside you have to fire Python up and write some code.

Unless you are really dealing with bigger datasets, there is not really a reason to use Pickle.

As a side note, even for big datasets (unless really really big), it is worthwhile to dump your data to both CSV and Pickle (even though online Pickle will be used for reading). This way you still get a readable version.

By the way, using the gzip module has better syntax for writing/reading compressed pickle files.