The Fire Department Calls for Service is a public
dataset from the city of San Francisco that logs all the responses to calls received by the fire department.
The data contains detailed information about the time interval of each call logged, therefore multiple records might be found for a call number.
Besides the time interval, each record contains call number, incident number, address, unit identifier, call type, call type group and disposition.
The dataset contains approximately 4,897,527 rows in total, therefore we will be working with a subset. Since our goal is to identify the incidents
happened during the wildfire season in California, we'll be looking at the calls made during the months of September to November of the past 5 years (2014-2018)
which comprises 378,816 rows. The dataset has 34 columns, we will consider some of them to create our prototypes and visualizations.
For more information about our choice of subset, see the Project Description section on the Home Page.
As the topic of our project is to look at the fire trends that may correlate to the months of September, October and November, we filtered the dataset to only look at those months for the years 2014 to 2018 (last 5 years). To reduce the size of the dataset even more, we also filtered only the columns we wanted to use, which were:
Even though our focus was only the months of September to November, we also wanted to give some context to our focus in relation to the overall data of those years. Instead of using all rows for each year, we calculated the number of rows per month for each year, and created a new dataset with those numbers. Apart from calculating total number of calls, we also calculated number of calls with specific characteristics such as calls with "Call Type Group = Fire" and calls with "Call Final Disposition = Fire".
The columns of the new dataset are: