Timeline for Forecasting: Linear vs. Exponential vs. ARIMA
Current License: CC BY-SA 4.0
7 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Sep 21, 2019 at 18:10 | comment | added | IrishStat | You say "In general, ARIMA should perform better than regression for forecasting time series data." I say autobox.com/pdfs/regvsbox-old.pdf | |
| Sep 20, 2019 at 19:24 | history | edited | user2974951 | CC BY-SA 4.0 |
edited body
|
| Sep 20, 2019 at 6:39 | comment | added | user2974951 | @MahmudulHasan Personally I don't see clear patterns in the data, the first plot looks like it might keep increasing, but the second looks like it could go either way. You may be better off using some simple predictions, such as the naive method. | |
| Sep 20, 2019 at 6:27 | comment | added | Mahmudul Hasan | So, what if I want to forecast this data, do you suggest I should solely rely on expert's opinion. On more thing, does 1st image looks have no trend as well? | |
| Sep 20, 2019 at 6:23 | comment | added | user2974951 | @MahmudulHasan Sure, something is better than nothing, however you still need a sensible something. For the second part, that depends on the data. If you really think that the previous data comes from a different process, then you can exclude it, since you are not really interested in. However, if this is about a period when something different happened, you can still keep it in the data and process it appropriately, such as using a dummy variable for war. | |
| Sep 20, 2019 at 6:20 | comment | added | Mahmudul Hasan | Don't you think it's better to have something than nothing in hand if you want to get a picture about future! May be it can be adjusted with expert's opinion. Another point is I had 36 years historical data, which represents some country employment. However, it faced war from 1985-1997 and data was very unusual that time. Before 1997 pattern and after 1997 pattern was very different. So, isn't it logical not use that part of data in forecasting? | |
| Sep 20, 2019 at 6:04 | history | answered | user2974951 | CC BY-SA 4.0 |