Conversation

I pulled every verified tweet from 2011 until today that mentioned shooting or shooter, removed sports references (basketball mainly), aggregated by hour and then checked all the spikes. Over 95% are US shootings. This is just preliminary for now. #dataviz #bigdata

JSON Baumgartner

@jasonbaumgartne

Aug 6, 2019

The largest spikes here are the Vegas shooting and the Sandy Hook school shooting. Basically every spike is some form of mass shooting in the US.

JSON Baumgartner

@jasonbaumgartne

Aug 6, 2019

At first, I went to aggregate the data by week, but there were too many shootings in the US to even aggregate by day, so I had to aggregate by hour to get a more accurate picture. This includes regional shootings, but spikes over 100 were mostly mass shootings. I had to remove ..

This Tweet was deleted by the Tweet author. Learn more

Replying to

and

No I have not -- what is it used for?

This Tweet was deleted by the Tweet author. Learn more

Replying to

and

You may find this tutorial helpful stumpy.readthedocs.io/en/latest/Tuto Note that .

@stumpy_dev

looks at a sliding window of observations rather than individual days/aggregates. So matrix profile anomalies might be revealed as a seven day window (high or low) that does not look like any other.