search

htm-community / NAB

The Numenta Anomaly Benchmark
GNU Affero General Public License v3.0
3 stars 3 forks source link

Plot cleanup #34

Closed breznak closed 4 years ago

breznak commented 4 years ago
breznak commented 4 years ago

@Zbysekz can I have your review on this PR? I've addressed the issue with TP,TN we've talked about. And some cleanup. I've saved the files without figures, so the ipynb is much smaller. (and easier to diff later changes)

Zbysekz commented 4 years ago

About the line in In[13] with comment "TODO is this branch needed?" i suggest to delete the condition and also the second branch with zero indexes index. Add this line instead: assert summaryDataFrame["Threshold"][detector_summary_row].size==1, "There is multiple records in summary file for one datafile!" I can't figure out why we would need to have multiple records per one datafile...

breznak commented 4 years ago

Two first plots uses same data file that user can change at the beggining ("detector_summary_row") but third plot uses different data file?

yes, the main modifications were to "Plot Results ..." file, while there was also a "Plot Data " which did just that, a single plot of the data file. I've decided to merge it just in a single file (?)

About the line in In[13] with comment "TODO is this branch needed?" i suggest to delete the condition and also the second branch with zero indexes index.[...] I can't figure out why we would need to have multiple records per one datafile...

I guess we can do that. I'd think it's for a compatibility with some older format, but currently NAB shouldn't change that much. I'll rm that

breznak commented 4 years ago

In the second plot ( In[6] ) i can't see any data for "standard score", just the legend.

thanks for spotting that. I've fixed the mistake I introduced and the data is being plotted. To be honest, I'm not sure about the meaning of that data and which block of code creates that. It's the 'NAB manually specified data of how anomaly score should go', and it shows -0.11 for non-anomaly etc. So we're again plotting that, but later we should look at how it's computed.

breznak commented 4 years ago

Thank you for your review, @Zbysekz . Please re-review when you have time, I've adressed the issues.