4 ms·
> It seems like this is a month to month number that fluctuates widely based on one-off events and things like when people typically pay their taxes. Even if i
by ineptech 5y ago
> It seems like this is a month to month number that fluctuates widely based on one-off events and things like when people typically pay their taxes.
Even if it varies a lot over the short-term, reaching positive numbers after a long period in the negatives would suggest an upward trend that isn't just noise.
- eurasiantiger 5y agoHow do you figure?
- PaywallBuster 5y agoe.g. tax receipts increasing over time so this years was bigger than the two years before (hence the biggest) which would suggest a positive/growing trend
- epivosism 5y agoimagine if the number were -X billion. As I increase X, your estimate should respond. Surely there is an X such that you would notice, even though it is admitted the number varies, right? i.e. even a one-time -100000 billion month would be noticeable in your framework? From there we show that -100 is not quite as bad, and +100 is better. If a new employee is an hour late the first day on the job, it doesn't prove they're not committed but it sure points that way. Every observation is evidence, raising or lowering your estimates of reality behind the observations.
- ineptech 5y agoYou can measure a long term trend, even if it's overwhelmed in the short term by much larger changes, by tracking local maxima/minima. Suppose that a circuit's voltage is either increasing or decreasing by 1V per hour, and your measurements ranging from 90V-110V. To determine the trend, you ignore the increase/decrease from one measurement to the next, and wait for it to hit 111 or 89.