0

Here's the problem : Each day, a number of virtual machines are created by users. I would like to forecast one day ahead how many virtual machines will be created during the next day. I have at my disposal a time-series representing the number of virtual machines created each day.

Time series

I cannot notice any particular trend, seasonality or correlation in this time-series, so I would like some advice on whether I should drop the idea of forecasting completely or if there is a forecasting model out there that could work well with my data.

  • Check if the time series is white noise, if it is then it's truly just made up of a random component – Malcolm May 21 '17 at 09:18
  • You can easily infer that the next day $7\pm 2$ machines will be created. As this is a statistic problem you will never be able to know what is the exact number. – N74 May 21 '17 at 09:42

0 Answers0