TL;DR: Don't give up! It gets better, and so will you!
I always try and obsessively look for the "best" or "proper" way to do something.
Don't. Do your personal best, and THEN look at ways to improve it, and in one facet at a time. Otherwise, you might not always understand why the "best" way is better. (You won't appreciate source control until you lose All The Things, but ... I'm sure you're already using Git. ;))
There are nigh-infinite blog posts about ways youcan do X better (unit test, standards, unicode compliance, date handling, names, design, CSS, layout, etc) -- don't worry about those yet.
=== Coding on your own ===
Consider writing code to solve programming puzzles, like Project Euler -- it can often start small, and then improve as you learn more.
Consider building a "simple" toy project in a web framework -- Django, Flask, etc. (I mention those as they are some Python frameworks.) There are often tutorials that run you through building something toy-scale. (Your own twitter, a TODO app, etc). Doing this will let you quickly build something with a simple UI, and simple guts, and then you can look into ways to improve it.
For example, I'd like to make a flash card game for my kids. (Real project, tired dad. ;)) I can start with some code that randomly generates problems, and shows them to a UI in a pretty way. (Large font, perhaps.) Later on I can improve it with things like letting people enter the answer, tracking how long it takes to answer, tracking scores over time (so that I can focus on the ones they have problems with), leaderboards, and custom decks. ALL of those are way out of scope of the initial project though.
=== Coding with others ===
If you'd like more of a challenge, or in some ways less, try being a contributor to a larger project, rather than trying to write something totally from scratch. This is really important -- most of us are not writing Large Software solo. I might write a handy tool solo, but most of my day-to-day work (and the environment in which I do most of my learning) is maintaining a larger codebase.
Existing projects already have a build framework, a code review protocol, RPC/API infrastructure, GUI, etc already done for you. They have a test suite that tells you if you broke something. Now, your improvements are things like, "Add an error message when this happens", or "Add support for a new API call...". It's MUCH easier to add something similar to what someone else already wrote (once you understand how it works by reading their code ;)), than it is to have to write all of that infrastructure yourself. There are also OTHER developers who have domain knowledge of the codebase, and that you can ask for help -- this is VERY useful and important for our learning as programmers.
=== Direction ===
I have no idea how to help you there. ;) It's easy to feel inadequate in this profession, as there are always smarter/faster people. Since what we can create is endless, it's hard to PICK something -- that's why I suggest doing a tutorial-based project in a web framework. You can then use that to focus on the Python side, or on the UI side, or as inspiration/guidance for making your own things.
Also, don't worry about looking at Stack Overflow. Definitely do that if you're stuck on something, or you forget whether something is possible ("Can I declare a dict plus it's members in Java the way I can in Python?"). It can at times be hard to tell when the answer you find on SO doesn't match the problem you really have, though, which takes practice.
I'd be happy to talk more via e-mail if you like. I've updated my user profile to include a poorly-obfuscated rendition of my gmail address. :-)