There is abstract of my last week analyses, on both intraday and intraweek merits.
ABSTRACT
Intra-day merits:
Notes:
- The part of the asbstract describes figures of intraday merits over the period from 19/2/2018 to 01/4/2019 (truncated dataset);
- Days from 24/1/2018 to 18/2/2018 truncated due to highly potential outliers; and days after 01/4/2019 truncated as well due to incomplete week (the 2019w14);
- Statistics presented in the post are for truncated dataset
(1) Potential outliers are days that have intraday total merits beyond 159 or 1127;
(2) Median of intraday merits over the period is 626;
(3) 50% of observed days have their intra-day merits range from 522 to 764 (the interquartile range);
(4) Friday [in GTM time] is the day over weeks has lowest intraday merits in terms of both median and mean, at 569, and 616, respectively.
(5) Monday [in GTM time] is the day over weeks has highest intraday merits in terms of mean, at 742.
(6) There is a flippening between Wednesday and Monday in terms of highest intra-day merits, with figures for Wednesday and Monday are 660 and 658, respectively;
(7) There are 26 potential outliers in total, and there is only three potential outlier days happened in early weeks of 2019, on 09/01/2019, 14/01/2019, and 27/3/2019, at 1161, 1127, and 1249, respectively.
<8> Minimum and maximum of intraday merits (full dataset) are 312 and 13018, on 11/2/2019 and 24/1/2018 respectively.
Intra-week merits:
Notes:
The part of the abstract use full dataset, only dropped last two days due to incomple week (2019w10).
(1) The median of intra-week merits is 4510;
(2) 50% of observed weeks (62 weeeks in total), have total merits in the range from 3854 to 5487 (the interquaritle range of intra-week merits);
(3) Minimum and maximum of intraweek merits are 3065 and 30949, in 2018w35, and 2018w4, respectively;
(4) Seven potential outliers [beyond 1405 or 7937], all of them occurred in the year 2018.
More details can be found there:
Time Series Analysis on Distributed Merits in the forum (daily, weekly, monthly)As we can see in the abstract, with truncated dataset, there are only three extremely potential outliers, that happened on 09/1/2019, 14/1/2019, and 27/3/2019, range from 1161 to 1249.