Evansville "Shots Reported" Data -- Part I

  1. Introduction

Evansville furnishes open data via a portal. There, a “shots reported” data set is maintained for different time periods. The data may be downloaded here. The time period examined is for one year. There were 864 observations with 15 variables. The time period spanned November 2, 2017, to October 3, 2018, or 335 days. Twenty-nine observations were eliminated as duplicates.

The data likely are generated from an emergency dispatch center and the individual calls are then coded as “shots reported.” So just because a shot was reported does not mean that it was exclusively the discharge of a gun. First, the call could originate as a result of a person believing a shot was fired, when in reality it was a similar loud noise. Second, multiple people might report the same gun shot and this would likely occur in places that are the most densely populated. Third, a caller might report a gun shot when one did not occur at all to, perhaps, draw a police response.

The data set was fun but quirky. Let’s start with some exploratory charts.

2. “Shots Reported” by the Hour

During the week, shots reported reach a peak around 4:00 p.m. in the afternoon and contradicted my expectation that they would peak later in the day. Instead, they seem to peak at a time when activity is at its highest: when school and work are concluding. On the weekends, the shots reported peak about an hour later at 5:00 p.m. It would be interesting to see whether dispatch call volume peaks about the same time.

3. “Shots Reported” by the Weekday

“Shots reported” by the weekday peaked on Saturdays and was at its lowest on Wednesdays.

4. “Shots Reported” by the Month

Because of a number of quirks about the data set, this particular chart should be viewed with suspicion. As noted above, the data set was actually for 11 months with October only having three days included. Another thing to note is that there was about a two week span in July where no gun shots were reported. This missing data will be explored in a future post.

Shots reported peak in May and December. My hypothesis is that both months include a legal holiday. In December, New Year’s Eve is celebrated and in May, Memorial Day is celebrated. July 4th did not have any shots reported and it skews the monthly totals badly.

Evansville Open Data Portal. Raw data should be viewed with suspicion for months of July and October.

5. Summary

The “shots reported” data set is an interesting metric that may have little to do with crime or violence. A shot is most likely to be reported around 4:00 p.m. on a Saturday in a month with a legal holiday. Grateful to Evansville for furnishing the data and hopeful more people will make use of it. There are some other interesting aspects to this data set—missing data for one—but I’ll leave it to a future post.