search

logv / snorkel

UI for interactive data analysis | https://snorkel.logv.org
https://fb.com/groups/snorkelsnorkelsnorkel
161 stars 21 forks source link

Errors with window time at small values #6

Closed rstecker closed 10 years ago

rstecker commented 11 years ago

Am trying to take frequent small samples to update a graph.

'end_date': '2013-09-20T22:50:00.000Z', 'end_str': 'Now', 'end_ms': 1379717400000, 'start_str': '-1 min', 'start_ms': 1379716800000, 'start_date': u'2013-09-20T22:40:00.000Z',

What's going on here? Why is it a 10 minute gap and not a 1 minute gap? I also know that I ran this on my machine at Fri Sep 20 15:45:04 2013 so ignoring any hour differences, why does it start at :40 and not :45?

okayzed commented 11 years ago

is the bucket for the graph set to 'auto'? you may need to manually specify the bucket size for the time series graph to 1 minute. you can look at the parsed params returned in the JSON to see how snorkel is interpreting the input.

On Fri, Sep 20, 2013 at 4:00 PM, rstecker notifications@github.com wrote:

Am trying to take frequent small samples to update a graph.

'end_date': '2013-09-20T22:50:00.000Z', 'end_str': 'Now', 'end_ms': 1379717400000, 'start_str': '-1 min', 'start_ms': 1379716800000, 'start_date': u'2013-09-20T22:40:00.000Z',

What's going on here? Why is it a 10 minute gap and not a 1 minute gap? I also know that I ran this on my machine at Fri Sep 20 15:45:04 2013 so ignoring any hour differences, why does it start at :40 and not :45?

— Reply to this email directly or view it on GitHubhttps://github.com/okayzed/snorkel/issues/6 .

rstecker commented 11 years ago

Errrr... Am not sure what you mean about bucket... Here's the whole parsed block:

{"limit":100,"dims":["event_type"],"view":"table","table":"redacted","sort_by":"count","stacking":"normal","start_ms":1379724900000,"end_ms":1379725800000,"end_str":"Now","start_str":"-1 min","start_date":"2013-09-21T00:55:00.000Z","end_date":"2013-09-21T01:10:00.000Z","hist_bucket":null,"time_bucket":300,"cols":[],"agg":"$count","cast_cols":[],"id":"api/1379725351"}

I see time_bucket is 300 but... I'm not really sure what that means...

okayzed commented 11 years ago

time_bucket 300 means to split the time series at intervals of 5 minutes (300 seconds). Because of bucketing issues, the time series will grab buckets from start_date - end_date, including the bucket that overlaps start_date and bucket that overlaps end_date, so that we get full buckets (instead of partially filled buckets at the end).

you can change time_bucket to 60 to turn on 1 minute intervals, which should give more precise results.

On Fri, Sep 20, 2013 at 6:04 PM, rstecker notifications@github.com wrote:

Errrr... Am not sure what you mean about bucket... Here's the whole parsed block:

{"limit":100,"dims":["event_type"],"view":"table","table":"redacted","sort_by":"count","stacking":"normal","start_ms":1379724900000,"end_ms":1379725800000,"end_str":"Now","start_str":"-1 min","start_date":"2013-09-21T00:55:00.000Z","end_date":"2013-09-21T01:10:00.000Z","hist_bucket":null,"time_bucket":300,"cols":[],"agg":"$count","cast_cols":[],"id":"api/1379725351"}

I see time_bucket is 300 but... I'm not really sure what that means...

— Reply to this email directly or view it on GitHubhttps://github.com/okayzed/snorkel/issues/6#issuecomment-24853200 .