_Pedestrian counts fromSan Diego’s smart streetlamps indicate that San Diegans responded strongly to stay at home orders, but the data has some weaknesses, and the reduction has several causes. _
In 2018, San Diego installed over 4,000 smart street lamps, one of the largest deployments of the CityIQ system in the world. These Internet-connected street lights give developers access to data on the intensity of pedestrians and traffic. But not anyone’s identity – the data is completely anonymous.
We’re about a month in to a crash course on data visualization; nearly every day you are seeing graphs of COVID-19 cases, hospitalizations and deaths, and there has never been a time when so many people know what a logarithmic axis is, or wish they did. Yet, there are some interesting features of the most common graphs that you may not be familiar with
This post will describe the shape of the most common curves, and explain how they are related and what they mean. Note that all of the plots in this post are based on data published by the County of San Diego, which the San Diego Regional Data Library has collected and packaged.
We are probably on the down-slope. It looks like San Diego has hit its peak in new coronavirus cases. This plot shows the daily increase or decrease in the number of confirmed coronavirus cases for San Diego County. In the ideal case, the full plot would look a lot like a Normal distribution ( aka bell curve ) so you can see that we’d just hit the peak a few days ago. From here out, we should see, each day, a smaller number of new cases than the day before.
The IHME projections for hospital utilization are probably the most popular data of the movement, just as the Financial Time charts were a few weeks ago. You’ve probably seen variations of this plot
everywhere online, and variations of it from The New York Times and NPR. You’ve also probably seen contrarians criticizing it for being unstable. Those contrarians aren’t entirely wrong; the error margins — the light purple band, for instance — are very large, representing a large uncertainty in the projections. That should not be surprising, since predicting the future is difficult.
Similar to the previous post about case growth rates in US states, growth rate are declining in most countries.
The global average for the decline in growth rates is about .9% per day. Extrapolating this decline would result in countries hitting zero new cases per day at about 40 days from when then hit 100 cases. China had a growth rate of .7% ad day 40, and less than .1% at day 52. A more accurate prediction model would be non linear, which is had to see in most of these plots, except for South Korea, which clearly shows the approach to zero.
Some US states, like California, have declining growth rates in COVID19 cases, others, like Massachusetts, are in trouble.
This chart shows, per state and for the US, the trend in the growth rates of COVID19 cases.
Each day, the number of cases is increasing, but by a smaller percentage each day. If today had 30% more cases (for instance) than yesterday, tomorrow may have only 28% more than today. That decrease means the growth will eventually be zero, then then we will have no more cases.
We are scraping the San Diego County coronavirus cases page to generate a historical dataset. See the data repository for a link to the file.
Do you know Biotech companies, non profits, or other organizations who are working on the front lines of the pandemic who need professional help with managing, cleaning or publishing data? The San Diego Regional Data Library and Civic Knowledge are offering pro-bono and low-cost help with pandemic-related small data projects. Some examples of the work that our professionals and volunteers can do include:
For an example of the kind of work we do, see the Data Library website and our data repository
Here is the video of my talk at the PyData Los Angeles 2019 conference, on the topic of “Tackling Homelessness with Open Data”. link to the presentation in multiple formats.
Here is a link to the presentation in multiple formats.
I’m presenting tomorrow at the PyData Los Angeles 2019 conference, on the topic of “Tackling Homelessness with Open Data”. I hope you can join us there, but if not, here is a link to the presentation in multiple formats.
If you are following along, you may be interested in our many blog posts on homelessness, including the example Colab notebook. You may also be interested in the Homelessness data collection.