Here is a file that simulates a Power Company metering every 15 minutes over a 7 day span for 1000 accounts. It is was a great way to illustrate clustering by similar power usage profiles. The file is self-extracting by an embedded script. After the script completes, there will two files that are joined virually by Account. There is a little GUI that allows you to change the parameter settings, how often the meters are sensored, how many days and how many accounts. Just be careful, you can end up with a giant file and there is a row counter that shows the row count before you commit. There are also a couple of embedded scripts for clustering and visualization to get you started. I recommend saving about 10 clusters and using local data filter on cluster. It does a nice job of separating the types of accounts that are built into the simulation. There are three elements that will cluster, the peak power, the range of power from min to max during the day, and the time of day that the peak occurs. One example of this is "Airports", these are set up to have an abnormally late peak, otherwise they would look like other account types.
