Hospital Access across Neighborhoods
These visualizations seek to view response times across different neighborhoods in SF, specifically looking at hospital transit times, to see if any patterns emerge.
Data Wrangling
To create these visualizations, I downloaded a subset of the original dataset, filtering down to 2018-2020 Call Dates and the Medic and Private Unit Types to focus specifically on hospital related response times. To create my prototypes in Tableau, this subset was further filtered to only include the data from 2019, which was aggregated and depicted as averages for the year.
I created calculated values to find the difference in time between:
- Recieved DtTm and Hospital DtTm
- Recieved DtTm and On Scene DtTm
- On Scene DtTm and Hospital DtTm
Due to errors in the original dataset, there were negative response times calculated, which were filtered out. The "None" neighborhood rows were also removed, as they did not contribute to the analysis of response times based on neighborhood.
The "Null" category of the Call Type Groups was included in the heatmap dataset. Given that this category of calls had hospital response times, it can be assumed that this category still demarcates significant emergencies occuring that just did not fit into any of the other categories, or were not marked with a category. There were enough data entries in this category that it is significant to include in the visualizations.
These edited Tableau datasets were exported to be used in these visualizations.
Background
Data Encoding
Heatmap: The heatmap compares rows of neighborhoods against columns marking specific Call Type Groups. The coloring marks the time in minutes from recieving a call to arrival at the hospital. Darker colors indicate longer time spans. The grey cells mark where data was missing from the original dataset. The heatmap was simply sorted alphabetically by neighborhood, as each column looked at a different set of data, and sorting by any one column's values would give unintended importance to that one Call Type Group in the visualization.
Bar & Line Chart: The chart has neighborhoods on the x-axis and minutes on the y-axis. The line denotes the time in minutes from recieving a call to arriving on scene. The bars denote the time in minutes from on scene to arriving at the hospital.
Interactivity
Hover over a specific bar on the Bar & Line chart in order to highlight it and see the specific times and neighborhood associated with that bar and that point on the line.
Hover over an individual cell on the heatmap to see the specific time associated with that neighborhood and call type group, as well as highlight the relevant neighborhood on the Bar & Line chart. This allows for comparisons of the different time intervals for a given neighborhood across both visualizations.
Visualizations
Hospital Related Response Time Averages by Neighborhood in 2019
Chaitrika Budamagunta
Fire Department Calls for Service
Source: https://data.sfgov.org/Public-Safety/Fire-Department-Calls-for-Service/nuek-vuh3
Conclusions
The overall theme of this project was to assess the effectiveness of the SF response system for the entire SF community, and hospital access is a critical part of evaluating whether or not the current system functions equally for all neighborhoods and populations. Using these visualizations, there are a couple different conclusions that can be drawn regarding this.
It can be seen that, in 2019, the overall average response time for life-threatening and non-life-threatening emergencies is fairly consistant across all neighborhoods. However, for fire related emergencies, certain neighborhoods face much longer overall times than others, namely: Castro/Upper Market, Lakeshore, Presidio, and Seacliff. This might suggest that the fire response teams assigned to these specific neighborhoods need extra resources or manpower to speed up the process.
Another interesting point that can be seen is the disparity in response times for South of Market specifically. This neighborhood had a drastically low average time between recieving a call and arriving on scene, but it had the highest average time between on scene to arriving at the hospital. Because of these disparate values, the average overall time across Call Type Groups is on the lower end of the heatmap scale, implying fast response times for the process. In truth, this neighborhood faces long transit times to get hospital treatment, and when it comes to medical care, a few minutes can be the difference between life and death.
Overall, it seems that, while the current response system functions fairly well for most neighborhoods, there are a handful of areas where it should be improved. Populations in certain neighborhoods face much longer wait times than others, lowering their chances at recieving the same important, possibly life-saving medical care as other neighborhoods. Assessing these specific areas and providing them more resources would help improve medical access for the entire SF community, and increase the effectiveness of the SF response system.
Attribution
Code examples used: