Open Data Preview Night - Fire Incidents

For a while now our office have been wondering how to engage folks when the city releases a new open dataset. Looking around at what works from other cities we noticed several have had success with data preview nights so we decided to give it a go. The basic idea is that you provide the dataset to interested individuals and have them "kick the tires" on the dataset.

Enter Open Chattanooga Hack Nights

Since the city was already providing space for Open Chattanooga's hack night at City Hall we decided to piggyback off the hack night for our first data preview night. Now, all we needed was a new open data dataset for the data preview. We went with the two related public safety datasets, 1) the fire incident dataset which contains one record per fire related incident and 2) the fire incident details dataset which contains multiple records for each fire related incident depending on the number of units responding. These two datasets contain a ton of great information about fire incidents in Chattanooga and were the perfect candidates for the data preview.

Civic Hackers...Meet the Fire Data

So right away those present identified several problems with the datasets. From fire codes missing, to missing addresses, to problems with the formatting of the response times, they were quick to hone in potential problems with the dataset that could confuse users of the data. We are currently working to address these problems which were not our radar before the preview night.

One of the most important parts of the dataset is the fire incident code. Each fire incident has a corresponding fire incident code that tells what type of incident occurred and there were several questions about what the various codes meant. So the Open Chattanooga members suggested we provide information about the data (also known as metadata) so users could get more details about the incident codes. As a result of that suggestion, we have attached a detailed overview of fire incident codes to each data's metadata section. Without their suggestion, this would not have been added to the dataset and many future users of the dataset would probably be just as confused as well. Another suggestion we followed was to also include a station location map (embedded below) on the open data portal.

Stranger Things

We did figure out several interesting things when looking at the datasets. For example, did you know most of the calls fire goes out to are not directly fire related? In August 2016, 94% of the fire incidents were coded as non-fire incidents. It surprised many folks at the meeting that the fire department responds to so many non-fire calls but since they are also dispatched on EMS calls they are often the first on scene for 911 EMS calls. But perhaps the most interesting finding of the night was which fire truck was the most unlucky (ie gets called on the most false alarms). Turns out that Q2 at Fire Station 1 on Main St has this dubious distinction.

Everyone agreed that the data preview model was a great way to engage folks with new open data sets. So what datasets would you like us to preview next? Click on a link below to suggest a dataset!