Shirley K Data


911 Call Project

This python project analyzes 911 call data from Kaggle. This data set is from Montgomery County, PA over the dates of December 10, 2015 to June 29, 2020.

The Jupyter notebook containing the code for this project can be found here.


These are the top 5 zip codes and top 5 townships for 911 calls in Montgomery County.

Zip Code Count
19401.0 44785
19464.0 43155
19403.0 34299
19446.0 31715
19406.0 22071
Township Count
LOWER MERION 54686
ABINGTON 39302
NORRISTOWN 36931
UPPER MERION 35412
CHELTENHAM 30044
 

There were 148 unique reasons why people dialed 911 in Montgomery County, PA during this period. The most frequently occurring reason was for EMS services.

 
Reason
EMS 326850
Traffic 227045
Fire 98797
 
p_p1_img1.png
 

The total 911 calls for each reason sorted by day of the week shows that the fewest calls occur on Sundays.

p_p1_img2.png

Fire and EMS are called rather consistently with a slight dip on Saturdays and Sundays. 911 is dialed for Traffic most on Fridays and least on Sundays. Traffic calls are more frequent during the work week and less frequent on Saturdays and Sundays.

 
p_p1_img3.png

This plot shows the total 911 calls for each department sorted by month. The Fire Department sees a peak in calls in June. Traffic calls decline from July to September and then rise steadily through the end of the year. EMS has a similar pattern, with calls slowing from roughly June to November and then more calls the rest of the year.

 
lat lng desc zip title timeStamp twp addr e Reason Hour Day of Week
Month
1 62336 62336 62336 55294 62336 62336 62312 62336 62336 62336 62336 62336
2 55427 55427 55427 48922 55427 55427 55405 55427 55427 55427 55427 55427
3 60027 60027 60027 53252 60027 60027 60001 60027 60027 60027 60027 60027
4 53671 53671 53671 47349 53671 53671 53655 53671 53671 53671 53671 53671
5 57509 57509 57509 50354 57509 57509 57474 57509 57509 57509 57509 57509
p_p1_img4.png

We see a drop in calls starting in June/July until November/December. We also see spikes around New Year, in March, and at the start of summer.

 
p_p1_img5.png

Here we attempt a linear fit, but as we can see, the variance is large and the error is extremely high except from roughly April to June. However, this very roughly shows that there is a rough declining trend showing fewer calls in the second half of the year and more calls in the first half.

 
p_p1_img6.png

Now we group by date and see that there are certain dates that experienced higher than average calls.

For example, there is a peak on March 2, 2018, which coincided with a Nor'easter that occured in that region and most likely created poor driving conditions and car accidents.

This happened again on November 15, 2018, where the region saw another heavy snowfall which caused many car accidents as well.

On June 3, 2020, there was another peak in calls caused by a tornado warning/high winds.

 
p_p1_img7.png

In line with the previous plot, we see spikes on the same dates as before, where harsh weather patterns caused poor driving conditions and a rise in traffic incidents.

 
p_p1_img8.png

We see that there were many calls for the Fire Department on March 2, 2018 during the Nor’easter and June 3, 2020 during the tornado warning, but not on November 15, 2018 during the heavy snowfall as we saw with Traffic and will see with EMS.

 
p_p1_img9.png

EMS accounts for most of the calls to 911 and is called for most consistently. Traffic calls average roughly around 100-150 per day in the date range specified. The Fire Department averages closer to 50 calls daily in the date range specified. EMS, in contrast, averages roughly 200 calls per day in the date range specified.

The peaks in data for EMS is less pronounced during the emergencies discussed for Traffic and Fire, but we can see there are still some interesting peaks that were less noticeable in the previous plots.

March 2, 2018 aligns with the Nor'easter as before, whereas the August 13, 2018 peak coincides with flooding in the area, and as before, there is a peak on June 3, 2020 during the tornado warning/high winds.

 
p_p1_img10.png

We can see some correlations between frequency of calls by day of the week and hour of the day.

The highest number of calls on weekdays between 2:00PM and 6:00PM, but this heatmap seems to indicate a higher occurence of calls during work week hours. The calls begin occurring later in the mornings on weekends, with Saturday seeing a higher volume of calls than on Sunday. We also see that there is a slight increase in calls later at night around midnight on Friday/Saturday and Saturday/Sunday.

 
p_p1_img11.png

We can see which days of the week and hours of the day are most closely correlated.

There is less activity at 3:00AM and 4:00AM fairly consistently. 3:00PM to 5:00PM on Fridays are particularly eventful, but 3:00PM to 5:00PM have an increased count in calls fairly consistently through the work week, perhaps due to rush hour. As before, we see higher rates of calls during the work week and during work hours. We can also note that lunch hour typically has more calls as well.

 
p_p1_img12.png

This heatmap shows correlations between frequency of calls by day of the week and month of the year.

Fridays in March seem to correlate to the highest volume of 911 calls, and Sundays in general, but more so in August and November, correlate to the lowest frequency of 911 calls. Again, we see that Saturdays and Sundays in general see fewer calls.

In the second half of the year, as seen in the first few plots, there are fewer calls than in the first half and last month of the year.

p_p1_img13.png

We can see which days of the week and months of the year are most closely correlated. It echos what we observed in the previous heat map: Fridays in March have a higher frequency of calls and Sundays in August and November have the lowest frequencies in calls.

 

In summary…

In Montgomery County, EMS is the most consistently and frequently required emergency service, followed by Traffic services.

Daily, most 911 calls occur during work hours.

Weekly, most calls occur during the work week and dip slightly over the weekend. There is also a noticeable peak in traffic incidents during Friday rush hour times and a noticeable dip on Sundays.

By month, we see a higher frequency of calls during the first half of the year and a slight lull in the second half of the year, except in December where calls increase. When we analyzed certain peaks, we saw that these calls corresponded to periods with inclement weather causing poor driving conditions. These may be more likely to occur during December and first few months of a given year.

Next
Next

Project Two: Finance Project