Energy Graph not scaling properly when switching between 5/15/30 mi

[CorneliusRox]

I like looking at the Energy Graph a lot. I feel like it helps me understand what impacts the range and what habits can net me some extra range.
One thing I don't like is when you switch between 5 mi, 15 mi, and 30 mi, the y-axis does not change it's scale. That means it shows -300 to 900 all the time. This would be fine, but it does scale the actual values, which means one of them are off.
If I'm at 400 Wh/mi at the 30 mi graph, it then shows >900 Wh/mi if I switch to the 5 mi graph.

Is this a bug that everyone else has noticed too?

 

[CorneliusRox]

Nobody?

 

[GtiMart]

I *think* the Y axis is fixed. The x axis will represent a different period, average values will probably change. Values over the max Y will just be cut to the max (or min). The actual consumption values are not scaled either, but since you show a longer distance at "30miles" than at 5, the 5 mile data gets squashed horizontally. You get less x pixels per mile, so less samples are shown per mile. What appeared as a wide peak plateau might now just be a small spike.

Edit: if you show us screenshots of 5 and 30, we might be able to better explain

 

[KenC]

Nobody?

Maybe some sample pics of what you mean by scaling the actual values?

 

[jdcollins5]

The number of sample points does not change. For 5 miles you have the most sample points per mile. For 15 miles you have less and at 30 miles you have even less.

 

[CorneliusRox]

Screenshots. Good call. And sorry if I explained it poorly.

Both images were taken back to back while sitting in my garage:

 

[GtiMart]

Thanks. Your explanation wasn't that bad but it was hard to believe Now that I see the pictures it's still hard to believe but at least I can't just dismiss it.
I went back and had a look at mine... although I'm in kilometers, I have the same thing. In my 10km graph I see I have a 1-kilometer spell under 0 (regen) but cannot see anything near that in the 25 and 50km graphs. Similarly, my wide peak around 500whkm in the 10km graph shows at around 300whkm in the 50km graph.
I could understand losing some peak values if they were over a very short period, due to varying samplings for the graph construction, smoothing etc. However I have wide enough "plateaus" that it cannot explain it.

I believe yours might be explained by local smoothing, when going from the 5 miles to the 30 miles graph. IF you don't want to see a very jagged graph with all the 800-900wh/m peaks along that 30 mile range, you might locally smooth those areas of the curve.

EDIT: Yeah, the more I think about it, the more I think that if they kept all the details of all samples like in the 5 mile graph, on the 30 mile graph you would get a very jagged line that would be impossible to make sense of. The (major) smoothing makes it so the graph gives a sense of how you consumed, it doesn't give the actual precise consumption at each point.

 

[CorneliusRox]

So smoothing out all the outliers... I do suppose that makes sense. Now I want to test it and see, but I'm not sure how. If I kept my foot into the floor for five miles... I'm sure I'd go to jail. ha ha

Any suggestions?

Anyways, thanks for taking a look!

 

[GtiMart]

I'm as curious as you now but I also wonder how to properly test. Do you have a mountain close by where the climb is a few good miles without any real flat or drops?
Alternatively, maybe generating a lot of wind resistance would do the trick. Or someone that does some track racing could give us some screenshots

 

[KenC]

I'd say the time increments change. On the 5mi chart, the time increments are correspondingly smaller, while on the 30mi chart, the time increments are larger. The smaller increments will generally have higher and lower peaks. The 30mi chart with larger increments, as has been noted, has the result of smoothing, as there's more averaging. Of course, if you do test it by keeping your foot floored for as long as whatever the increment size is for the 30mi chart, you might get the same peak.

It's like looking at different maps of the same location. If you change the scale, what looks jagged at a close-up scale, will look less jagged at a zoomed-out scale.