Extra Credits is a video series about game design topics, it's pretty good.
One of the episodes was about scoping projects. Too many people start stuff, never finish it to the point where you feel you can show it to someone else. That's bad. A couple of ideas are important here:
Rule 1: get a playable demo out as quickly as possible. As a rule of thumb, game design books suggest getting a playable build no later than 20% through your planned development period. e.g. for a game that takes two weeks to complete, you want a playable build within 2 days. For a 1 week project, you want a working build on the first day. It doesn't have to do much, just proof of concept: it should have placeholders for all the main game elements you plan to flesh out for the rest of the week. Avoid feature creep here: if you didn't plan for it, don't include it unless you've completed all other features within the allotted time.
Rule 2: have a well-defined "end product" you want to make, and an end date to make it by. If you fail to meet that date, then start something else, but smaller scope the next time. e.g. make 1 game per week, no matter what. If you don't complete it on time, do not get bogged down: scope out a new project straight away. This is good experience. Finish up old projects in the quickest way possible when you have spare time (a playable thing you can show to someone is always the goal).
It doesn't have to be a game, per se, just something you could happily demonstrate to someone else, e.g. even a D&D treasure table generator could be a good first project. You then take all the modules you've made, and use them for spare parts when scoping the next project, e.g.: "take my treasure generator, and build a simple exploration game where you fight goblins and amass treasure, selling it at a town" would be a 1-week project that could follow a 1-week project where you learned to make a treasure generator.
If you're completely new to programming, start small, but still stick to the time management ideas. e.g. make a rock-paper-scissors game against the computer. It's not much, but if you've never coded before, it will take you a while to learn what you need to code that.
Once you've made a whole pile of 1-week projects, then start thinking bigger. The experience will give you a good grasp of what you're able to achieve in a set time, which will gradually increase as you have many more code parts to play with. Working through a bunch of classic genres is one of the best ways to start. Tic Tac Toe, Tetris, Pong, Space Invaders, Breakout: these are the level of project which are good for beginners.