I was actually going to make a thread about these haha. They're used enough everywhere now that it's worth going over what they are.
So RAPM charts give the results of regression analyses performed with target variables of the things we see on the chart - goals for, expected goals for, corsi for, expected goals against, corsi against. These are all stats that most people are happy to look at individually, so it's nice to have them all in one chart. Further, the regression does something a lot better than just comparing the raw numbers on a chart - regression analyses can be used to isolate the impact that one variable (player) has on these "target variables" or stats. They build a matrix that contains every NHL player and goes through every shift (defined as a period of time in which no changes are made and ES hockey is happening (or maybe 5v5)) and keeps track of these stats. the regression analysis gives a coefficient for each variable (player) and ultimately tells you the impact THEY have on their team's GF, xGF, CF, CA, xGA. Isolated completely from other players, both teammates and opposition. So when you look at that chart, roughly, you can say that "this already takes into account the fact that Mitts got to play in the offensive zone a lot against other team's bottom players, and that Larsson has to face tough opponents in heavily skewed minutes." It's certainly far from perfect like any hockey model,, but no statistical model ever claims to be perfect - just useful. And regression is an objectively useful statistical technique that tells you things. So it's not just that "ROR or Bergeron have a good +/-" - it's that "ROR or Bergeron by themselves contribute a LOT to their team's goals for and preventing goals against compared to the average NHL skater."
Keep in mind that the stat which is the best at predicting future team and individual level goals-for is expected goals, and the stat which performs the best at goals against is CA. So it's got the two best predictor stats on there.
And then it displays the results in terms of standard devation from the average NHL player. So that means a few things.
The size of the bars should not be viewed as uniformly increasing or decreasing in impact on the stat. If you're at +1 STDV, your impact on the stat is better than all but ~15% of NHL skaters. If you're at +2, it's better than all but ~2.2% of NHL skaters. If you're at -1 you're in the bottom 15th percentile, and if you're at -2 you're worse than all but ~2.2% of skaters in terms of impact on that stat. And with a sample size of however many NHL skaters there are, this chart then tells us that Sobotka's impact on expected goals, because it's so hard for him to create or use space to set teammates up or get shots on net himself, is only better than maybe 2-3 skaters to play some amount of minutes in the NHL this season. He's approaching the bottom 0.1 percentile. (and we used him as our 2C by minutes played and 4th-most-used player for a significant chunk of the season - but anyway)
So if the bar is between 0 and +/- 1, you cover a lot more ground in rankings with small movements up or down the bar than you do towards the extremes. So Vlad is nowhere near as negatively-impactful (words) on those defensive metrics - perhaps he's in the bottom 20%, but nowhere near the bottom 0.1%. And Larry's impact on defensive metrics is top 10%/top 3% of all NHL skaters.
ie, ~68% of all NHL skaters fall in this circle, so stuff i there is a bit more muddled than stuff on the edges.
So to answer your question, I like the charts because of everything they bring to the table, and they make it so I have fewer awful data tables to look at, and does a lot of the comparison work for you. It's far from perfect of course, but even if I had the TIME to gather the context for any player comparison in the league I wanted to make, watching game after game to discern usage and impacts myself, I'd be just as flawed as the regression is in my own analysis. They're quite useful, do a lot of work for you, and like any other stat, you should focus on looking at extremes and ask yourself what hockey characteristic you think may be contributing to or influencing the result.
Slap a few of these charts down, add a dash of your favorite counting stats, pull up some film and you've got yourself a neat little hockey analysis