Timeline for Confidence Intervals for Mean Difference between Actuals and Forecast
Current License: CC BY-SA 3.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 26, 2018 at 15:51 | vote | accept | stats-hb | ||
| Jan 15, 2018 at 19:32 | comment | added | stats-hb | not exactly - I do not know, if my t-test-approach for the confidence intervals is valdid in a time series context (with potential autocorrelations in the data) and when one of the time series is not "manifes" but an estimation itself (with its own uncertainty) | |
| Jan 15, 2018 at 19:25 | comment | added | jbowman | Oh... so the problem is that you don't know how to get the pointwise confidence intervals from the forecast object? | |
| Jan 15, 2018 at 19:23 | comment | added | stats-hb | I actually produced the "bad" forecast deliberately to show a clear distinction between the forecast and the actuals, because I am mostly interested in learning about the confidence intervals for their mean differences. (I probably should have thought of a more realistic example, but this was the first, that came to my mind) | |
| Jan 14, 2018 at 23:17 | answer | added | jbowman | timeline score: 2 | |
| Jan 14, 2018 at 22:45 | comment | added | jbowman | You put an "N" in for the seasonal component, so naturally your prediction doesn't have one. Since virtually all the time series structure is due to the predictable annual variation, you're missing out on it. | |
| Jan 14, 2018 at 20:43 | history | edited | stats-hb | CC BY-SA 3.0 |
deleted 2 characters in body
|
| Jan 14, 2018 at 17:40 | history | asked | stats-hb | CC BY-SA 3.0 |