Fair enough. I wouldn't have the slightest guess on average trips per licensed driver, but I would think that using driver instead of car would be more accurate.

Lol. Well if you don’t want to reach your potential then you’re exactly right. So If by realistic you mean what you’re willing to work on, then yes. And I’m not knocking that at all. I personally have never taken up a sport with the goal to be as mediocre as possible but I’ve known those who are happy to just play regardless of how they do. 

Greetings, I thought I'd circle back on this topic and let everyone know that I've decided to stick with my current coach.  My current coach has really taken me a long way. Not only that, but I think it's "easier" (is that the right word) to stick with one voice. I hit the ball brilliantly on the range the last couple weeks before getting injured, but it didn't translate to the course (actually quite the opposite). I think some of that was due to me being kind of stuck between the two worlds. While injured I've been diving into my video and my swing was never better that it was while under his instruction. So, back to it.  Since I've been injured I haven't had a chance to work on much of anything new. However, he and I are putting together my plan for the next 6 months. We will use all my data including Arccos, as well as launch data from Trackman. Going to analyze weaknesses, firm up strengths and get to the plan. It should be a fun off season.  Thanks to all who responded to this thread.  One more thing: Really looking forward to getting back to the SuperSpeed training. 

Give it another 3-5 years and I might feel the same. Some tech within that umbrella is ready to go out and be a commercial success, but I've done enough work with some of the latest and greatest for machine learning algorithms to know I don't want it touching anything critical safety related just yet. Computer vision is something that many people have a pretty good handle on and there's fresh data coming in often enough that 1-3 missed detections aren't going to cause any noticeable affects. Machine learning is quite a bit more difficult and most practical applications using it are currently "stuck" in that 80-90% of the way there phase that's good enough to show real promise but not good enough to release to the public. Funny enough it's a lot easier to teach an AI to play a video game than it is to teach them to do something like control a temperature. It wasn't all that far off it turns out, the real number was 0.00029 compared to my estimate of 0.00039. The reason I said 3 trips per registered car on average is because while there are a lot of cars that sit (multiple cars for one driver, or the driver isn't using it that day) there are also plenty of cars that are moving all day possibly with multiple different drivers. Figured it probably averaged out somewhere in that 3-4 trips per car range and it looks like that's correct (since 4 trips per car would give a probability of 0.00028. 4 trips per car per day would've been the best guess it seems.

I will accept "easier" for many but for me both more realistic & easier!  + Why make things more difficult than they need be?