This may be too much work, so I understand if no one tackles the problem... I'm doing an analysis of the power usage of a building. Sheet 2 of the attached spreadsheet shows typical data (we have 4 years of it). The top graph is the kWh used every 30 minutes of every day in January, 2003. The second graph is the temperature at those times. As you can see, there is a correlation between temperature and usage (the building has electric heat). For example, the light and dark blue diamonds for 1/1 and 1/10 show the highest temperature and lowest usage. What I would like to do is filter-out this correlation. This would leave me with a usage profile that reflects occupancy and pretty much nothing else. All of the lines for kWh would lay right on top of each other most of the time. Sheet 1 shows some of my efforts to this point. The graph seems to show a corellation between kWh and temp (though, admittedly, not super-strong). Of note, there is something called an "equilibrium temperature." Its basically the temperature at which no heating or air conditioning is required. At this point, the usage will start going up again. So this would serve as a baseline - a minimum usage - for building the corellation. It appears to be at about 50 degrees. Now, it looks like it should be possible to just construct this line and subtract it from the usage data. But it doesn't appear to be working. Any help would be greatly appreciated.