Timeline for Problem in discrete valued time series forecasting
Current License: CC BY-SA 3.0
20 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 13, 2017 at 12:44 | history | edited | CommunityBot |
replaced http://stats.stackexchange.com/ with https://stats.stackexchange.com/
|
|
| Mar 25, 2012 at 14:54 | history | notice removed | CommunityBot | ||
| Mar 25, 2012 at 14:54 | history | bounty ended | CommunityBot | ||
| Mar 20, 2012 at 15:16 | vote | accept | nb1 | ||
| Mar 19, 2012 at 18:11 | vote | accept | nb1 | ||
| Mar 19, 2012 at 18:11 | |||||
| Mar 19, 2012 at 18:10 | comment | added | nb1 | that is indeed interesting. I will try digging deeper. | |
| Mar 18, 2012 at 23:46 | comment | added | IrishStat | Nikhil Then you must have been amazed and astonished at my findings '...suggest that the data might be daily as the ACF of lag 7 suggests structure" . Perhaps there is something systematic in the "arrival of these states". A clear consistent effect was found at day7,day21,day28 with day14 being flagged as an unusual day breaking the pattern. | |
| Mar 18, 2012 at 18:47 | comment | added | nb1 | daily data it is. | |
| Mar 18, 2012 at 11:50 | comment | added | IrishStat | @NikhilBellarykar What is the interval between the readings ? Is it daily data or is it just random intervals between the readings ? | |
| Mar 17, 2012 at 22:34 | comment | added | nb1 | It is not ordinal data but purely class labels. | |
| Mar 17, 2012 at 18:40 | comment | added | IrishStat | @Conjgate They appear to be ordinal, If they were purely attributes/classes/labels , ARIMA processes would seem to be inapplicable. | |
| Mar 17, 2012 at 17:26 | comment | added | conjugateprior | So these are absolutely not ordinal data? They do seem to behave that way... | |
| Mar 17, 2012 at 16:30 | answer | added | conjugateprior | timeline score: 6 | |
| Mar 17, 2012 at 15:49 | comment | added | IrishStat | I believe that I have provided you with a state-of-the-art solution with sufficent detail as no data-based model (compared to a pre-specified assumed model) is currently available. The model I suggested below using your data deals with predicting (low count) discrete valued time series using the history of such a series and any level shifts/local time trends/seasonal pulses/pulses/changes in parameters/changes in variance deemed necessary and sufficient. | |
| Mar 17, 2012 at 14:38 | history | tweeted | twitter.com/#!/StackStats/status/181026828943765504 | ||
| Mar 17, 2012 at 13:38 | history | notice added | nb1 | Improve details | |
| Mar 17, 2012 at 13:38 | history | bounty started | nb1 | ||
| Mar 13, 2012 at 0:32 | answer | added | IrishStat | timeline score: 1 | |
| Mar 12, 2012 at 18:10 | history | edited | user88 | CC BY-SA 3.0 |
edited title
|
| Mar 12, 2012 at 16:01 | history | asked | nb1 | CC BY-SA 3.0 |