How to achieve overlay graf with daily total, average per day for a week and standard deviation?

[fatsug]

I'm trying to use tstats to calculate the daily total number of events for an index per day for one week. Then calculate an averade per day for the entire week, as well as upper and lower bounds +/- 1 standard deviation.

Feels like I can get each individual thing to work, either the bar chart with the daily count, OR the statistics. Though I want them all in a single figure.

This is my base search:

| tstats count as daily_events where index="index" earliest=-7d latest=-1d by _time span=1d 
# | timechart span=1d sum(daily_events) as daily_events

Which seems to work just fine by itself, with or without a following "timechart"

Now I figure I'd do something like this:

| stats sum(daily_events) as total_events stdev(daily_events) as std_dev 
| eval avg_events=round(total_events/7, 0)
| eval upper = total_events + std_dev 
| eval lower = total_events - std_dev

 Which by itself seems to work, though I now loose the results from the base search.

So how can I combine these things (can I combine?) into something close to what I want to achieve? I'm attaching a crude scetch for clarity.

It feels like I somehow need to add three columns just replicating the upper, lower and av_events values for every "daily_events" entry, or something. I just can figure it out. Any feedback is mych appreciated.

[yuanliu · Accepted solution]

You must be looking for eventstats (in lieu of stats)

| tstats count as daily_events where index="index" earliest=-7d latest=-1d by _time span=1d
| eventstats sum(daily_events) as total_events stdev(daily_events) as std_dev
| eval avg_events=round(total_events/7, 0)
| eval upper = total_events + std_dev 
| eval lower = total_events - std_dev

[fatsug]

I was just about to post myself but glad you beat me to the punch, totally forgot about aventstats!

Thank you