hybrid plateau

[bwilson4web]

Originally Posted by MechanicalRandall

What is the unlabeled x-axis of that graph?

There were ~187 samples in the largest data set. So I shifted all curves to put the middle at 100. By putting all mid-points at 100, it is easier to see how the mileage distributions fit within the distribution.

One reason for doing this instead of a traditional distribution curve (aka., a Gaussian curve) is the problem of small count samples. If we tried to do a distribution curve, the processing would distort the value to the point of not being decipherable.

Bob Wilson

[MechanicalRandall]

Originally Posted by bwilson4web

There were ~187 samples in the largest data set. So I shifted all curves to put the middle at 100. By putting all mid-points at 100, it is easier to see how the mileage distributions fit within the distribution.

I think I understand the point of the graph, just not the actual arrangement of the data. I understand that you shifted the curves, but what actually makes these curves? How is the data arranged along the x-axis and what are the two colored rectangles for?

Sorry if I'm looking too far into this.

[dcb]

B sorted the respondents for a vehicle by mpg, plotted the mpg for each respondent, then centered the plot on 100. Repeat for several vehicles. The x graph is basically 180 buckets for individual responses for a vehicle. with vehicles superimposed on each other.

[jamesqf]

I think you've got a basic problem in your "hybrid plateau" idea, because (other than the couple of 2nd gen Insights), all your data is from what's basically the same car, the Prius. Even the Insights are a car built to be similar to the Prius.

Plot a 1st gen Insight on the same graph, or a PHEV conversion, and you can see ways to make improvements.

[NeilBlanchard]

Here's another data set that may or may not correlate with your chart:

http://futurevehicletechnologies.com...ss_Release.pdf

[mpgx2]

I'm new to all this.. studying and playing around with mods..

trying to get serious but reading up a lot first..
--

Me and some friends joke about hybrids. Of course we
do it out of jealousy. That's all it is.

Anyway, our (cynical) theory on hybrids is that the electric
motor is only powerful enough to keep the motor turning over
under no load and to drive the aircon.

When the car rolls down a hill, the petrol motor is actually
off but it appears that it is on.

Rolling down hills with motor off saves a lot of fuel. I know,
that's what I do all the time in my non-hybrid and it saves
25% of fuel on my particular to and from daily drive.

My friend in germany always harps on about theory.

He says that 35% of fuel is wasted in heat and just
goes straight out the radiator. Even modern engines
have no heat recovery mechanisms.

Then, another 35% is wasted internally in cooling the
motor's combustion chamber.

So, 70% of the hybrid's fuel is still perphaps being
wasted.

I read this as those guys making hybrids have jobs
for life.. because there's still a long long way for
them to go.

[bwilson4web]

Originally Posted by MechanicalRandall

I think I understand the point of the graph, just not the actual arrangement of the data. I understand that you shifted the curves, but what actually makes these curves? How is the data arranged along the x-axis and what are the two colored rectangles for?

Sorry if I'm looking too far into this.

Once I have the MPG data for each vehicle, I sort them individually so they are in order from lowest to highest.

I add an index column from 1 to 200. Then I shift each vehicle so the middle of their values corresponds to 100. Call it the median, which it is. So the medians of each vehicle cross at 100. Then I plot all vehicles.

Now the two rectangles highlight:

• linear range - those that appear to fall on a line
• exponential - the extreme mileage cases that fall outside of 'normal.'

Those are not strict math definitions but 'eyeballed' to make sense.

Bob Wilson

[dcb]

Originally Posted by bwilson4web

...
Those are not strict math definitions but 'eyeballed' to make sense.

Ah, ok. The bottom gap should probably be closer since the mpg scale is not linear.