Timeline for What can you do with 'crazy' data?
Current License: CC BY-SA 3.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 9, 2011 at 21:16 | comment | added | Wayne | @Iterator let us continue this discussion in chat | |
| Aug 9, 2011 at 21:15 | comment | added | Wayne | To be honest, I'm considering approaching the utility and the state commission about having the utility's statisticians make data available from which we might make reasonable models. They wouldn't have to make the models and they could have disclaimers so they're not sued if my forecast turns out to be wrong... My fear is that on the billing side, they really do ad lib most of the time, knowing they can provisionally raise rates and refund them later, or can retroactively ask to recover losses they didn't properly anticipate. | |
| Aug 9, 2011 at 20:52 | comment | added | Iterator | I understand. I have some ideas on hedging with energy futures contracts, but that will likely lead to sorrow. :) It would be a form of insurance to cap upper costs. On the other hand, a few utilities have shown interest in giving more consistent energy price caps under some circumstances. Well, I guess it's worth asking for better updates from the utilities. It's not like they don't have actuaries and statisticians. :) | |
| Aug 9, 2011 at 20:17 | comment | added | Wayne | The one good thing is that the bill is for the Association itself, not individual units. (I.e. our hallway lights, ventilation, and air conditioning, our elevators, pumps, and other infrastructure.) I am calculating my model's accuracy by cross-validation and my usage model is around +/- 6%, while my cost models are +/- 10-20% (or even 30%), hence my frustration. No way we can say to our fellow owners, "Hey, we may charge $25,000 extra this year for electricity, to play it safe, and so may have a boatload of extra cash at the end of the year." Think torches and pitchforks. | |
| Aug 9, 2011 at 20:03 | comment | added | Iterator | Also, I'm not sure that a range of $20-40K (+/- 10-20%) is that unreasonable. 1. Consider overall #s of people - although it's a slight stretch, an apt bldg could easily have 10% vacancy or more during certain stretch of the economy. 2. Sums can be affected by outliers. It may only take a few extreme people to push total usage up a bit. | |
| Aug 9, 2011 at 19:57 | comment | added | Iterator | Well, you're a condo association which is presumably a member of a larger statewide condo organization. :) Think in terms of scale, and then you're talking clout. Just ping the largest representative group you can, and see if they can get the data for you. It never hurts to see who can pull some levers for data. A few breadcrumbs such as a little report that they can send their members could go a long way in convincing them to check on this. | |
| Aug 9, 2011 at 18:03 | comment | added | Wayne | Thanks! In terms of being large enough to get more info from the utility, we're a condo association so we're medium-sized and fall into that awkward gap of: large enough that we're talking a fair amount of money, yet not large enough that the utility especially cares about us and not as financially flexible as a company. We're talking roughly \$200,000 for electricity annually, and the cost models I've created (using lmer) have 95% credible intervals (by simulation) of \$20,000-40,000 which too large to budget against. | |
| Aug 9, 2011 at 17:23 | history | answered | Iterator | CC BY-SA 3.0 |