I am trying to figure out an approach for calculating the probability of a renter making > 1 booking. I would prefer to use Python for this project. I have the following columns ready: State, Category (fun, adventure, etc.), returning_renter? (TRUE if >= 1 booking).
What would be the most efficient approach for this project? Are there any particular Python libraries you would recommend?
You could trivially prototype your solution in Orange (it's a Python library that uses scikit-learn under the hood). It also has a neat GUI.
But with only two features (state and renting_reason), I'm afraid your models might not be as precise as you'd want them to be, unless you have lots of examples to learn from.
The "standard" approach, when using python, would be data manipulation with the numpy library. If you have more complext needs pandas library is also a very good idea.
In order to train a predictive model I would highly recommend scikit-learn. It offers a wide range of machine learning algorithms and they have a very good online documentation and examples.
Depending on your needs there may be more efficient approaches but as a rule of a thumb your implementation using the aforementioned libraries will be good enough.
If you are not dealing with a big dataset I would suggest using pickle in order to save intermediate python objects instead of having to recalculate them again.
