Waiting4M3
Active Member
I wonder for the blue line, if it may be better to use median, or use avg of the middle portion of the VIN distribution, say between 25-75 (or 10-90) percentiles, thus excluding the skewing effect of lower VINs from rework, which belongs to a different statistical distribution from the main body.As promised, I reworked my chart on VIN assignments now we have a week more data. The data range is from March 8th till today, 1 day shy of 4 weeks. View attachment 291528
For each day, I graph the maximum VIN assigned that day (green), the minimum VIN assigned (red) and the average of all VINs assigned (blue). In addition I added the trend lines for each series.
The green line (maximum VIN assigned) tells us something about line speed : how fast do model 3s roll of the line. That rate over the given period is from 9400 to 15300, or just over 1500/week. The assumption is that each top VIN represents a car that goes through quality control exceptionally fast and needs no rework. Ie, that car is nearly immediatly after production complete available for customer assignment. R square is .9 or a good fit.
The blue line (average VIN assigned) tells us something about production speed where production includes regular quality control and rework. This rate is from 8000 to 12200 or 1100/week. R square is .85 or a reasonable good fit
The red line (lowest VIN assigned) tells us something about quality control issues. How slow is Tesla in reworking cars that don't pass quality control on the one hand. The spread of this line with the green line tells us if Tesla has sufficient capacity to rework cars as they roll of the line. This rate goes from 6300 to 8500 or just below 600/week. R square is .3 so not really a good fit.
My conclusion : the run up to 2000/week was recently enough that it doesn't yet show in the VIN assignment numbers. Rework and quality control is growing but because production rate is growing as well, it's too early to tell if this is sign of a significant backlog, especially with such a weak trendline.
The reason I think this is that if a car is reworked, it's taking a spot in the production line and preventing a new VIN from getting started, so the impact of rework is already reflected in the new VIN #s that are starting in the line. If we then average the rework VIN with the new VINs, we may be double counting its effect.