MF_ReadingByHourAccumulating

Fact

This fact table is structured a lot like MF_ReadingByHourSnapshot, except that it contains data that may be aggregated using SUM, in addition to MIN, MAX, AVG, etc. The data here, that is, is fully aggregatable.

For example, this table might hold data like 'System available time' (in seconds). This can be aggregated from hour to hour, day to day, to a total figure (giving us cumulative availability). Cubes based on this table should offer a percentage measure for time-based values calculated as (time value) / (total elapsed time). This allows us to see things like what percentage of the time a system was available.

A figure like 'Average power output,' however, does not belong here, since it cannot be aggregated. An average generally can't be summed. Likewise 'Power output,' if it indicates a wattage currently being produced, measured in, say, kW (kilowatts), cannot be summed. All you can do is average it over some period of time, or take its MIN, MAX, etc., and from there calculate kWh, kilowatt-hours.

Everything here is allocated down (or aggregated up) to the hourly level, so that everything is comparable, at least along the time dimension(s). Hours - the basic unit of temporal analysis here - must begin at minute 0 second 0 and end at minute 59, second 59. Hence there is no need to store end dates/times, since all end dates will be the same as start dates, and all end times will equal the start time + 59 minutes, 59 seconds. Rounding will be needed if a measurement's original grain is lower than per second. Links to MD_Time are only to rows where MD_Time_Second = 0.

Unlike MF_ReadingByHourSnapshot, this table has no fields for MIN, MAX, or AVG reading, although those can certainly be calculated, and may in fact be useful. They would be calculated for a grain higher than 'hour' like 'day' or 'year.'