1) I would hope we can agree that by "analysis" we mean to process data to provide information in a useful format. As such, we may collect a mass of data (rpm, laptimes, and temperature in this case) and select the data that is most useful to us. For instance, we might decide to analyse engine performance only when water temperature exceeds 50 deg C. By applying such criteria to select a mass of data, we have effectively "analysed" all the data, but determined that only a small sample set is useful to us. In this case, we could easily have set the logger to only start recording when the engine temp exceeded 50 deg C, but because we have no cost/storage limitation, we simply collect it all and filter out afterwards. One day we may seek to go back to the data and look again, but this time at the whole range of temperatures.
2) Your assuming that it isn't processed because we have too much, but that is not the case. It isn't processed because our assumptions define criteria that determine what data would be useful to us. Again, the use of criteria to limit the scope of a sample set is an analysis itself.
3) Regarding your "engine pulses" etc etc. We don't do it because of cost/availability limitations. Using your McLaren example, do you truely believe McLaren analyse every byte of data? Of course they don't. All the data is processed in some way, but an engineer will not analyse every byte. This is no different to selecting a sample of useful data and no different to what PaulM described. I could understand your point if PaulM were talking about a random sample of data, but I don't believe he is.
4) Choosing not to collect data. Again, this is not because of some inherent problem processing and analysing the data, it's because of cost/storage limitations.
To summarise, where cost/storage etc limitations do not apply, you cannot have too much data. You simply develop a set of assumptions that determine criteria for selecting a sample set of data that is useful for answering the "questions" at hand.
|
|