Closed smindinvern closed 6 years ago
Any thoughts, anyone?
I'd like to compare your code and mine for efficiency. If yours is faster than mine, I would definitely go with yours.
sigh... Now that i'm actually using the same data file as you are in the brown
branch, the results line up. And, unsurprisingly, the implementation in brown
using pivot
is about 2x faster than the implementation in this branch. The only thing I wonder about is how well this method will scale, since it's resampling the data to 1Hz, which, for some number of years of data is... a lot of seconds... Probably still fine though for this use-case.
Add functions for computing per-period (most likely hour) utilization for each machine, and merging that all into a single DataFrame, with columns representing individual computers.
This implementation is horribly slow.