We have a unique connection with our daily commute routine, be it taxis, buses, subways, or our personal cars and bikes. Unwantedly, we had to change that routine amidst the virus outbreak in different ways. Although the pandemic has brought upon some challenges, we do have some ways to deal with it! In this blog, we will dive deep into what’s happening right now in the city of New York and find if our daily commute is safe or not.
At Locale, we have built a console which shows you the latest risk scores for various NYC stations right now amidst the Covid-19 pandemic. You can explore it here:
In case you are using your phone, we recommend using your PC to explore the link here: https://nyc-subway.locale.ai/analysis/public/Static-Analysis-Z3GC32D6UJYR57SDRQ5D
1. How safe is your daily commute in NYC?
With over 383,770 coronavirus cases and 30,244 deaths, New York City falls unders the world’s most badly hit cities by coronavirus. Over 8 million New Yorkers residing over 302 square miles of landmass make the role of public transportation very important when it comes to daily commute. In fact, it is distinguished from other U.S. cities for its low personal automobile ownership and its significantly high use of public transportation!
NYC Subways have been serving as the major transit choice for New Yorkers. The Metropolitan Transportation Authority or MTA, which operates most of NYC’s transit systems, reports a massive 1.7 billion subway rides as of 2013! That’s huge when comparing it to a 803 million figure for the city buses.
Now coming to the current situation, the subway usage has obviously decreased due to coronavirus. But, what about people like doctors, essential services staff and other people, who still can’t afford a personal commute and rely on the subways? They are the ones most at risk but they have limited choices.
In the upcoming sections, we will learn how we can leverage data to make the daily commute of New Yorkers even safer!
2. Letting data do the talking
The New York City Health Department published the zip code level covid cases daily here. Let’s try to plot this on a map and see how cases have increased across NYC over time.
Here’s how it looks as of now:
These are the top 5 worst hit areas in terms of total positive cases:
Corona (quite a coincidence!), with 241, 940 cases, seems to be the most badly hit area right now, followed by Bronx, Queens, Brooklyn and Bronx.
Now, how do we know about the commuters daily movement patterns in the NYC Subway? MTA provides this data on a weekly basis which can be found here. This gives us the total number of entries and exits at each station at a 4 hour interval. We can get the total footfall at each station by summing entries and exits. Let’s plot some of the busiest stations.
We can also find top 5 busiest stations by finding average footfall. Plotting the total footfall at these over time, we get this:
With an average of ~66k people travelling daily, 151 — 145st station is the busiest, followed by 59st stations and 23st stations.
But, which stations have seen the maximum weekly increase in commuters after the covid outbreak? For this, let us restrict to the data after March 22, as it is that time around which the NYC governor asked all non-essential businesses to close and residents to stay at home.
To achieve this, we first find the total weekly footfall. After this, the change in footfall at each station can be found by simply subtracting the footfall at first week and the latest week.
The results are quite interesting! Out of total 240 stations, 20 have seen an weekly average of increase in traffic from 23rd March onwards. At these stations, roughly a thousand more people have travelled after the lock-down started.
“Of these 20, the top 3 stations which have seen increase are Junction Blvd, Bedford Ave and Longwood ave.”
You can see the increase in footfall for all of these 20 stations using this bar chart.
For the majority of the stations, there has been a decrease in the number of people using these subways, which is natural. Now, let us see how we can relate this to the coronavirus outbreak to find out risky stations.
You can explore the Locale.ai NYC Console here.
3. Which stations are risky?
With the combined knowledge of coronavirus cases in an area and the total footfall at a subway station, we could associate a risk factor with each subway station.
For the sake of simplicity, let us consider risk at a station to be proportional to the total coronavirus positive cases in the zip code which it lies in and the total number of travellers at that station at any given day.
We consider which zip code the station lies in for the total number of coronavirus cases. Although this is not a very good way to get this number for a station, we have to use zip limited by the data granularity available. You can find out more about why zip codes don’t do well in one of our blogs:
So, we can say that the product of total footfall and total positive cases at a station give its associated risk. Since these numbers are on different scales, we normalize them before multiplication.
So, our station risk formula looks something like this:
where NORM is min-max normalised value in time window t.
After some geospatial heavy lifting, we get the station risk on a day. We can see a visualisation of the same here:
From this, we can actually see what are the top 5 risky stations over time.
If we consider today, the most risky stations are 451, Junction Blvd and 449 on 111st Street.
These stations have the maximum virus outbreak in their neighbourhoods as well as maximum increase in the traffic, which has only been increasing since March 22.
You can find the code to the above analysis here: https://colab.research.google.com/drive/1vFLzK__popWXO6Kr-SUTlP5-cBbadqB4?usp=sharing
Similar to the station analyses, Locale.ai provides a way for businesses to track their static assets in real-time. For a company, static locations are entities of their business that don’t move. For example, restaurants, vehicle stations, warehouses, POIs etc. Static locations can affect the behaviour of demand and supply.
Hence, it is important to analyze how they perform to take decisions such as:
- Closing of stations depending on bookings, searches & idle time spent
- Improving efficiency of restaurants depending on prep time
- Provisioning or redistributing supply at warehouses based on where people drop off while ordering
- Opening up new stores, ATMs or branches depending on the demand (searches, installs)
If you are a hyperlocal on-demand company, chances are these static locations are a primary aspect of your business and you need to take different decisions such as closing them, opening them or improving their operations.