I think there's an in-between that's missing here. Folks who design electronic circuits don't typically compose a circuit directly from the properties of the parts. Instead, there are a lot of slightly higher level design elements, kind of like
algorithms in programming, that you combine to make something work. Examples are the many basic op amp circuits, lower level circuits such as transistor gain stages, and so forth. Studying enough of those circuits, you begin to recognize them in schematics, and this gives you a clue as to what the design is doing. I'm not sure that there's a good way around spending a lot of time memorizing those patterns.
There are also simplified operational models of some components such as transistors and op amps.
Assembling kits is a valuable first step, in that it gets your hands working. Being a bit more confident about your construction practices helps when you later try to make your own stuff, because you don't have to wonder if something like a solder joint is working. The more you can trust that your build is modeled by your schematic, the better a chance of locating the place where it isn't. That's your failure point.
I was a physics major in college, but a year of electronics was part of our curriculum. It was taught from The Art of Electronics, first edition. We were also required to buy the National Semiconductor Linear Applications Handbook, and a book on a mainstream IC logic family, which was TTL at the time. That stuff is all available online. I was fascinated with this stuff, so I read the applications handbook from cover to cover. IC application notes are still a good thing to study.
You can take another route. There's gobs of stuff you can do with a microcontroller board and pre-made peripherals such as sensors and actuators. It won't turn you into an analog jock, but it's a legitimate design method and might suit your fancy.