Boston Fire Incident Reporting, 2012 to 2025

Provider
City of Boston (Boston Fire Department)
License
Open Data Commons Public Domain Dedication and License (PDDL) 1.0
Format
Parquet (in sampled files)
Stored at
Amazon S3, s3://scigantic-boston-open-data/fire-incidents (us-east-1)
Access
Free and public. No account with the data provider is needed.
Category
Urban & Civic
Source
https://data.boston.gov/dataset/fire-incident-reporting

About Boston Fire Incident Reporting, 2012 to 2025

Every incident the Boston Fire Department responded to and reported under the National Fire Incident Reporting System, published for 2012 through March 2025: 641,957 rows across three DataStore-loaded incident files plus two code lookup tables. One row is one incident (or one exposure of a multi-exposure incident, indexed by exposure_number). Each carries the incident number, alarm date and time, the NFIRS incident type code and its description, estimated property loss and content loss in dollars, the fire district and city section, the neighborhood and ZIP, the NFIRS property use code and description, and the address decomposed into street number, prefix, name, type and suffix with cross-street components. The two lookup resources are the reason this dataset is usable without external reference material: a 188-row incident type code list and a 154-row property use code list that decode the NFIRS numeric codes in place. Most incidents are not fires. The highest-volume category in the main file by a wide margin is code 553 'Public service' with 106,036 responses, followed by 745 'Alarm system activation, no fire, unintentional' (39,563), 600 'Good intent call, Other' (37,281) and 714 'Central station, malicious false alarm' (34,403); the largest genuine fire category is 113 'Cooking fire, confined to container' at 32,102. So a count of rows is a count of responses, not of fires, and any fire-rate analysis has to filter to the 100-series fire codes. That distinction is what makes the dataset valuable for studying response burden and false-alarm load by neighborhood and property type as well as actual fire incidence and loss. Two structural facts are documented in the schema card. The 2012 and 2013 files use Title Case column names with spaces ('Alarm Date', 'Incident Type', 'Estimated Property Loss') while the main 2014-onward file uses snake_case ('alarm_date', 'incident_type', 'estimated_property_loss'), so concatenating the three requires an explicit rename map. And the older files write the literal string 'NULL' into the neighborhood column and pad city_section with spaces, so null handling and string comparison both need normalising. The main file ends 10 March 2025, so this series is not current to today. Published by the Boston Fire Department under PDDL 1.0. This archive is served from a Scigantic mirror rather than live from the portal, which is straightforward here because the series has stopped: the main file's last alarm_date is 10 March 2025. s3://scigantic-boston-open-data/fire-incidents/ carries a byte-verbatim copy under raw/, including both code lookup tables, alongside a Snappy Parquet copy under parquet/, FUSE-mounted read-only at ~/archive.

How to access Boston Fire Incident Reporting, 2012 to 2025 in Python

import boto3
from botocore import UNSIGNED
from botocore.client import Config
s3 = boto3.client('s3', region_name='us-east-1', config=Config(signature_version=UNSIGNED))
s3.list_objects_v2(Bucket='scigantic-boston-open-data', Prefix='fire-incidents/', Delimiter='/')

Lists the files with anonymous access. Runs on any machine, with no cloud account.

Data structure

File types in sampled files: Parquet

Files

Example file: fire-incidents/parquet/2012-bostonfireincidentopendata.parquet

Sample notebook

Sample notebook for Boston Fire Incident Reporting, 2012 to 2025: worked code with the output it printed.

Analyze this dataset without downloading it

On Scigantic, Boston Fire Incident Reporting, 2012 to 2025 opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Urban & Civic datasets