I’m one of those people who say don’t start a project by buying parts, Murphy sez you will need different parts.
OTOH at some point in learning enough code to DO THINGS well, one needs to buy lots of things, and storage boxes to keep them in. I like the compartmented tray boxes with lids that Harbor Freight (me no connection). You need a multimeter and if possible a temperature-controlled solder station.
And when you need to save up, work on your code skills. Minimum is Wokwi Simulator or a board and PC, you can learn how to run smooth multi-part automation with void loop() cycling at over 50KHz (loop counter function shows count/sec) while doing light work.
Once you make 2 things work “at the same time”, you can add functions made in other sketches… as long as they don’t block, the lesson to make 2 things etc. An Uno with circuit extensions can run scores of mundane tasks, each developed in its own little sketch. You can do this with just a board and a PC and time spent.
I can turn any sketch that uses delay() into one that doesn’t and gives back change… delay(1) wastes 16000 cycles, ie LOTS!
I’ll have to see my old undelay examples if I have them, replacing delay() with millis() timers takes 2 code skills but other than that(!) it’s cut and dry. You can automate with response begin times under 50 micros, except on contact switches cause they bounce.
This is all stuff you can learn with just a board while saving up to buy the parts you will be sure of first.
Here is project: fill a case with sensors, etc, that you made work and have code for so that you could use any in future projects with confidence of smooth integration at run time. If you don’t know a wide array of parts you might miss better choices.