1
$\begingroup$

I want to build a gauge graph where I can show if the current energy usage is high or low when compared to a previous period. The comparison would be shown as a percentage. So if the current usage is normal compared to a previous time set, the gauge graph is on 50%.

But, how to calculate this, without taking the high peaks that sometimes occur?

I could take the highest data amount in a time set and compare this to the current usage. So, for example, if I use 500W now and the highest peak in a time set is 800W the usage now is $\frac{500}{800} = 62.5\%$. But this is subject to distortion if I get a high peak for a few minutes because I turned on a heavy machine, and thus the percentage isn't valid anymore.

I was thinking about not comparing the current usage with the highest peak, but with the third quartile. What do you think of it?

$\endgroup$

1 Answer 1

1
$\begingroup$

If your readings are roughly normal, you can use mean and standard deviation as the benchmarks. That is, put the mean in the middle and highlight one or more standard deviations above and below it. Control Charts rely on the same concept, with more complicated limits.

If your readings are skewed, you might consider a transformation, such as plotting the quantile of each reading instead of the raw reading. That gives a linear scale with the median in the middle, but it may be too hard to interpret depending on your audience.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.