Hello,
I have the following SPL command :
|tstats count where index=main host IN (H1,H2) by host, _time span=1h | predict "count" as prediction algorithm=LLP holdback=168 future_timespan=240 period=168 upper90=upper90 lower90=lower90 | `forecastviz(240, 168, "count", 90)` |
It makes predictions about the count of a host
It outputs an array of the type :
host | _time | count | lower90(prediction) | prediction | upper90(prediction)
H1 | 2021-07-10 00:00 | 6170671 | 2494994.26372 | 6170671.0 | 9846347.73628
H1 | 2021-07-10 01:00 | 6231397 | 2456899.6988 | 6231397.0 | 10005894.3012
.
.
.
H2 | 2021-07-10 05:00 | 5216984 | 1722288.55477 | 5216984.0 | 8711679.44523
H2 | 2021-07-10 06:00 | 5297360 | 1979214.14187 | 5297360.0 | 8615505.85813
.
.
.
I would like to calculate linear regressions for each host in this table with the MLTK macro
`regressionstatistics("count", prediction)`
to outpuy an array of type :
host | rSquared | RMSE
H1 | 0.8042 | 1195199.83
H2 | 0.7842 | 1126684.87
I can't do it.
Could you help me?
Thanks in advance
Sincerely
M.
