Issue in temperature, day time and solar production ??
This post was updated on .
It looks like there are issues in temperature and time-of-day data available in boot files of 2018 trial_1. The temperatures are low in the day and high in the night. I randomly checked a couple of boot files on some random days (of the 14 day boot period), this issue is consistent. I can understand if such temperatures are high in the night on occasions and seeing a consistent pattern does not seem right. For your reference, I have two figures from Powertac_2018_trail1_1.xml (1st boot file of trial 1 of 2018). The first image shows the general trend for all 14 days of the boot file and the second image picks day 2 to illustrate the issue. On x-axis is the time of the day; y-axis is temperature. Convention used is time slot 24 is 12.00 hrs midnight (this convention is consistent with solar production data) .
We also looked at the solar production guy (specifically Middle_Income_2). They produce more during daytime and nothing in the night (that is ok); but they seem to have inverse correlation with temperature.
The index of the lists below is time starting from mid-night.
Re: Issue in temperature, day time and solar production ??
Thanks a lot John. Will wait for Govert to take a look at this.
I am sure you understand that if an approach relies on some correlation statistics between day time, temperature and production, then on games with such boot files, the approach may throw erroneous results not because of the problem in the approach but there is a timezone problem somewhere in the boot data.