Elio Crowdfunding

[bowers baldwin]

 

[bowers baldwin]

 

[AriLea]

145% by next Monday seems in scope. I'll post the chart tomorrow or Friday and we'll be more sure.

 

[bowers baldwin]

 

[Ekh]

I'm saying again that nice as it look, exceptions to the trend line should be greeted with suspicion. Ari, can you make a control chart out of your run chart? That would be even more informative. That's the way to know if things are stable, and to identify when they're not.

You probably know about this, but if you don't, here's a how-to using Excel.

 

[AriLea]

Well, I didn't before, but that video does make it fairly clear how standard deviation applies to quality control analysis. You know, you never claim any technical expertise, but you definitely have detailed insights.

The problem I've always had with standard deviation for this purpose is that without selective application, it's blind to the composite nature of real world elements. But really that's the purpose of SD anyway, to evaluate the data quality when you have no knowledge of the peripheral 'affectors'. (or don't want to be distracted by them) i.e, it's mathematical bound guess-timation. It's most useful for removing errant data input, and also identifying the highest risk parts of a chart.

In our case here for the blue line, there are simply corrections made, (if accurately corrected) that tends to simply displace the 'amplitudes'. Therefore, smoothing or rounding the curve is better than eliminating 'accelerations' sourcing from the red chart. For the red line that's not as true. But time displacements are mixed in there as well so eliminating is also problematic. In this case I would average between every four days, to get a realistic picture that would be more useful for future trending.

Both standard deviation and my smoothing tend to bury and hide the influences of errant events, which is kind of necessary to derive a trend. In both cases you could compare the final charts with the original ones, and the plottings that moved the most are suspect events. Or, these could be compared to real world events to identify 'affectors'.

Anyway, long story made short, I'm on the fence about that, so I'll just update the current charts. My bad! (next one probably tomorrow)

 

[Ekh]

Well, I didn't before, but that video does make it fairly clear how standard deviation applies to quality control analysis. You know, you never claim any technical expertise, but you definitely have detailed insights.

The problem I've always had with standard deviation for this purpose is that without selective application, it's blind to the composite nature of real world elements. But really that's the purpose of SD anyway, to evaluate the data quality when you have no knowledge of the peripheral 'affectors'. (or don't want to be distracted by them) i.e, it's mathematical bound guess-timation. It's most useful for removing errant data input, and also identifying the highest risk parts of a chart.

In our case here for the blue line, there are simply corrections made, (if accurately corrected) that tends to simply displace the 'amplitudes'. Therefore, smoothing or rounding the curve is better than eliminating 'accelerations' sourcing from the red chart. For the red line that's not as true. But time displacements are mixed in there as well so eliminating is also problematic. In this case I would average between every four days, to get a realistic picture that would be more useful for future trending.

Both standard deviation and my smoothing tend to bury and hide the influences of errant events, which is kind of necessary to derive a trend. In both cases you could compare the final charts with the original ones, and the plottings that moved the most are suspect events. Or, these could be compared to real world events to identify 'affectors'.

Anyway, long story made short, I'm on the fence about that, so I'll just update the current charts. My bad! (next one probably tomorrow)

I got involved with Total Quality 20 years ago, did the Deming 3-day, but couldn't do that stuff now to save my life. One of the things control charts give you is "flags" for an out-of-control process. 7 consecutive points in one direction, or 7 in a row above or below the line, and you have to ask what's going on. Single events outside the lines tend to be "special cause" events -- if traffic deaths in Ohio are 'under control" with 25 to 30 per week, and suddenly there's a 65 week, your control chart will show that -- and investigation shows that a school bus went over a bridge. Not something you can predict or even have to do something about, even though it's a horrific example. "Common cause" variations are much harder to identify and correct. Those would be trolls, IMO.

 

[AriLea]

The funding numbers are starting to run again. This maybe in response to the P5 images. What we want now is a run in the Elio Reservations at the $1k-all-in level.

 

[Ekh]

The funding numbers are starting to run again. This maybe in response to the P5 images. What we want now is a run in the Elio Reservations at the $1k-all-in level.

Maybe someone has a plan to help that along!

 

[bowers baldwin]