I am very often asked to recommend how to get started in data science, especially what should be that best first project. Actually, it is quite clear: have a business goal, start easy, do not aim too high at first, and avoid all complexity while valuing transparency. I have just published an essay which explains the types of projects you should avoid—and the ones I recommend for your first projects. Read it here:
Good luck with data science—and see you at one of my week-long classes where I teach those practical aspects in person, while you build your very first models.
Rafal