Data Study on NYC Fire Incident and False Alarm
The skills the author demoed here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.
If you live in New York City, as I do, you have likely noticed that the sound of sirens cuts through the background city noises round the clock. These sirens made me wonder, how many are real fire incidents, and how many are false alarms? To determine the answer, I created an app using R Shiny to show the data breakdown and enable people to easily read and understand what accounts for so many sirens in the city.
You can see my project and code links below:
Two of the datasets I used were from NYC OpenData. In the fire response data, it records every incident responded by NYC firehouse from 2013 to 2017. Inside the dataset, there is a section for incident code, which distinguishes each type of incident. I grouped incidents by fire incidents, false alarm, and other incidents.
In this section of the project, I observed the number of false alarm, fire incident and other incidents, and find the highest incident for each New York City Borough.
The column graph above shows out of the five boroughs, Brooklyn had the highest rate of fire incidents in the years between 2013 and 2017. They amounted to 55,672. One of the reasons Brooklyn had the highest rate is because Brooklyn has the highest population and the highest number of housing unit in New York City.
This graph shows Manhattan had the highest rate of the false alarms-- 115,842 incidents between 2013 and 2017.. Manhattan has the highest population density in New York City with 72,033 per person per square mile. Human or fire system failure may cause a false alarm. Human failure means there may be an eyewitness report that mistook something for a fire or an accidental activation of a fire alarm.
Frequency in a Day
This graph point out the frequency false alarm and fire incident occur in a 24 hour period. The highest false alarm rate occurs between 1pm and 5pm, and the lowest range happened from 1am to 6 am. This is because people are less active at night and more likely to be sleeping than cooking or actively using electricity. On the other hand, fire incident increased from lunch to dinner time. These period have the highest fire incident rate due to the need for fire or heat for cooking. Still most events turn out to be false alarms as The graph below shows.
Frequency in a Month
We’ve established that time of day plays a role in fire alarm incidents. the question is: what about time of year? This monthly frequency graph shows the highest range of false alarm happened from June to October, which would be the warmest time of year. On the other hand, the highest range of actual fire incidents occurred from October to April, which is the coldest time of year. Due to the cold weather, people massively use heaters to increase the indoor temperature. Thus, some old building with old wire might contribute to fires caused by the overuse electricity or shorts in electric heaters.
As the graphs above showed, we can distinguish the fire incident and false alarm happened in each borough by population and population density. We can also see the role time plays, both in terms of season and time of day.
The analysis would be more thorough if it also offers a visualization of geographic analysis.Using zip code or borough to segment fire incident or false alarm in the map by a different color can make the project has more information about the incident based on geographic location. I would use other coding languages like Java or Python to create these graphs and add it into the project.