Hi, sorry for reviving this very old thread. I just got another question about this - but this time the questions was not how a monthly value can be 85% if all weeks are complete, but how it can be 125%. This led me to think/investigate a bit further, and unless I miss something, isn’t the solution to this simply to make sure the calculation of "expected" reports in a period is always based on the same number of periods that was included for the "actual". (This would apply to aggregation of completeness over time, not indicators in general).
A concrete example from June 2015: 5 weeks have start date in June and contribute to the numerator/"actual reports" for a weekly dataset aggregated to monthly completeness, but only 4 weeks are included in the denominator/"expected reports". Reporting rate summary does this differently, so here June 2015 has 4 weeks for both expected/actual. (But reporting rate summary messes up July 2015, including only 3 weeks in the "expected reports". I.e. it seems it only includes weeks where both start and and dates fall within the month in the numerator, and gets the denominator based on something else (assuming there are 4 weeks in a month perhaps?). Thanks, Olav > 19. jul. 2013 kl. 19.39 skrev Lars Helge Øverland <larshe...@gmail.com>: > > Hi Olav, > > we simply use the start date of a week and see which month it falls within. > This applies to all period types. > > This has the consequence, which you point out, that for a month you the > system might include a week which is not currently finished, and hence cannot > have any data, making the completeness statistic funny. > > There might be better ways of handling this, like not including periods in > aggregation periods which are not complete. On the other hand, a week must > contribute to a month in any case - it is also not that good to include weeks > which starts in a previous month. > > Aggregating weeks to months is in any case dodgy business and I am not aware > of any standardized rules for this (please educate me if someone knows > better). > > cheers > > Lars > > > > > > > > On Fri, Jul 19, 2013 at 2:30 PM, Olav Poppe <olav.po...@me.com > <mailto:olav.po...@me.com>> wrote: > Hi, > I have a question regarding how the monthly completeness figures for weekly > datasets are counted. Some users have complained that even in months where > all weekly reports are complete, the monthly figure is 85%. > > What are the criteria for the week to fall within a month for the "target" > and "complete" respectively. For example, is week 31 (29 July to 4 Aug) > counted as July or August? Is it the same for the "target" and "actual" in > completeness (if not, that could explain how monthly completeness is not 100% > when all individual weeks are complete)? > > Olav > _______________________________________________ > Mailing list: https://launchpad.net/~dhis2-devs > <https://launchpad.net/~dhis2-devs> > Post to : dhis2-devs@lists.launchpad.net > <mailto:dhis2-devs@lists.launchpad.net> > Unsubscribe : https://launchpad.net/~dhis2-devs > <https://launchpad.net/~dhis2-devs> > More help : https://help.launchpad.net/ListHelp > <https://help.launchpad.net/ListHelp> >
_______________________________________________ Mailing list: https://launchpad.net/~dhis2-devs Post to : dhis2-devs@lists.launchpad.net Unsubscribe : https://launchpad.net/~dhis2-devs More help : https://help.launchpad.net/ListHelp
