About

This project used data collected and published by the City of San Francisco about the time and nature of each call that came in during the December of 2016. Visit SF OpenData for our dataset. This dataet has 27,768 observations on 34 different variables. Our visualizations explore 3 dimenions: the Battalion that responded to the event (Battalion), and the times ambulances picked someone up (Transport.DtTm) and arrived at the hospital (Hospital.DtTm).

The line and bar charts are pretty straightforward. The times were reported with accuracy down to the second, but we have binned them by the minute (rounding up at the 31st second) to increase effectiveness.

Note: Not all calls to the SFFD resulted in an ambulance call.

Tableau Prototype

D3

This first chart is an area graph with a callout focusing on a subset of the data. The x-axis is the duration of the ride from the scene to the hospital. This chart includes negative values in the data. Since these values don't make physical sense, they are a sign that something else could be going on. There's no way to tell with the information we have what that could be, or if it is important. It could be that fire fighters simply forget to push some first button until minutes after they've hit the second. Maybe this is just random error- or maybe it's because they're tired. Should we be worried? We can look to the next chart to find out.

This second bar chart does not encode number of records as height. In this chart, the height of the bar is proportional to the number of times a battalion reported a negative (impossible) time divided by the total number of calls that Battalion responded to, and the color is proportional to the number of calls that battalion responded to in total. Our motivation for visualizaing this calculation was to get a sense of whether or not some Battalions might be struggling more than others. From our exploratory analysis of the other fields in the data, we did not find and compelling correlates with the entries of negative times. The way this chart shows the tendency of a battalion to make record-keeping mistakes with height, and encodes the total number of calls that battalion responded to with blue allows us to see that Battalion 5 has seemingly the worst performance in this metric, but the color of the bar reminds us to take that value with a grain of salt. If there were an anomaly in the data, and Battalion 5 really was experiencing a problem, the sign of that problem would be a tall blue bar. The bars are organized from left to right, from most confident to least confident. The average rate of "mistakes" was 0.35%, and is labeled in the graphic.

Team

Click me to view my personal web

My name is Yuanyuan Ruan. I live in San Francisco. I am a senior student in data science major in Univeristy of San Francisco. I am a dream chaser, who enjoy life and challenges.