Skip to main content

Best Bin Placement: using open data to optimise public bin placement in Leeds.

How many public bins does Leeds need and where should we put them?

We've been trying to answer that question using two datasets,

  1. The location of every premises licensed to sell food in Leeds (as a proxy for the creation of public waste).
  2. The location of every bin in Leeds (as a proxy for the collection of waste).

There was lots of work to do before we could get started. We've already blogged about some of it. Our tool is now ready for you to try .

Mapping the existing bins

First we needed to know where all of the public bins are in Leeds.

We chose Open Street Map (OSM) as a database to store bin locations so that anyone can contribute and improve our dataset. And we worked hard to synchronise Leeds City Council's data on bin locations with OSM . As a result, we think that Leeds is the world's best city for open data on bins .

Estimating waste generation

We used premises licensed to sell food as a proxy for public waste creation. Our data on these premises comes from the Food Standards Agency (FSA) who publish data on food safety inspections for the whole of the UK. We extracted data for the whole of the UK and we've shared the code and the dataset on GitHub for anyone to use in their projects.

We had planned to consider just takeaways and sandwich shops as generators of waste, but we found that the classifications of premise type are too poor for that to be useful .

Best Bin Placement: two datasets in one tool

We've built our Best Bin Placement tool to bring these two datasets together. This lets us see the current gaps in bin placements and simulate the addition of new bins. This way we can start to answer the question we started with. Where should we put public bins so they collect as much of Leeds' public waste as possible?

What are the problems?

Building anything new is hard. We encountered a lot of problems along the way.

One of the biggest problems we found was that even with the best open data on bins in the world, data on where bins are in Leeds was incomplete. An example is Cardigan Fields Leisure and Entertainment on Kirkstall Road. This is a collection of entertainment (cinema, bowling, etc...), fast food, and restaurants surrounding a large car park about 2 miles West of Leeds City Centre.

The concentration of places licensed to sell food makes the location a hotspot of waste generation. And because there are no bins at the location on open street map it shows up as a prime candidate for extra bins.

Private sites often manage their own bins and these are rarely on Open Street Map.

But a quick look on Google Maps streetview shows that this is not a priority place for more bins. There are already plenty of public bins. But because the site is privately managed and the bins were not part of Leeds City Council's open dataset on public bins they don't show up on OSM.

Google streetview lets us see that bins exist even if missing from OSM.

Many sites where waste is managed internally show the same pattern. The Merrion Centre, Leeds Markets, Pinnacle, and Trinity Leeds all show up as places requiring more bins when in fact the bins that currently exist just need mapping.

The solution to this problem is to add these missing bins to OSM. This can be done directly using the OSM interface, or using the tool we have created just for that . Until that is done, some candidates for extra bins in our tool will be false positives.

Another place which the data shows needs more bins is Leeds train station. This is unsurprising. Bins in UK railway station are rare due to the risk of terrorism. It requires an understanding of UK culture and politics to explain why this problem is hard to solve.

So where should the bins go?

Understanding how our tool works lets us consider its limitations. I've already explained the problems with missing data, but this is just one of many limitations. Here are three more,

  • We do not consider the role of dogs in creating waste, so we miss bins that are required in parks.
  • We do not consider that different types of food premises generate different types of waste, so we over or under estimate the number of bins required in different areas.
  • We do not filter food premises by type (focusing just on sandwich shops and takeaways for example) because the FSA data is not good enough to let us do that.

But despite these limitations there is still value in what we've done.

A big finding is that the areas most likely to be underserved by public bins are in Leeds City Centre. Of these places many can be discounted as places that require new bins; they show up only because their bins are not mapped. But there are some that deserve more attention.

One such area is Call Lane. The road hosts many food premises and a number of cafes that sell food and drinks to take away. The narrow road is heavily used by buses and taxis which means that there are few bins. It looks like this in Best Bin Placement.

Call Lane has a large number of food premises but few public bins.

Just three additional bins provides very significant improvement.

Adding just three bins significantly reduces uncollected waste in our simulation of Call Lane.

Another genuine hotspot for public waste is the new development between Whitehall Road and Wellington Street to the West of Leeds Station.

The area between Whitehall Road and Wellington street is another area of unserved demand for public bins.

Placing two extra bins reduces the problem.

Two extra bins greatly reduce the problem.

Put together the placement of five new bins is obvious on a more zoomed-out look at both interventions.

The modelled effect of five extra bins on unserved demand for public bins in Leeds. Better data will be required to place bins optimally as many of the remaining hot spots contain bins that are not on OSM.

We've been building, releasing, listening to feedback, and improving this tool for a few months now. One of the biggest things we've learned is that there are a lot of bins -- currently about 4000 in Leeds alone -- and people don't want to place new ones individually. So each bin that's added to the map counts up to ten times what a normal bin does.

Other requests we've had include,

  • Using the tool in reverse to see which bins can be removed with least impact.
  • Creating a special version of the tool just for recycling bins.
  • Automating the creation of an optimal distribution of bins.

What's next?

Of the many requests we've had we're working on two for now, expanding to more places and considering more types of waste.

More places

The FSA data that we use as a proxy for waste generation is available for the whole of the UK. OSM provides a free database for the whole world that people could add bins to. So our tool could work with almost no modification for the whole of the UK.

The problem is that most of the UK's bins are not on OSM. We hope that the tool we've built for Leeds will help places make the case for more places to collect data on where their public bins are and publish them in an open way via OSM. If they do, our tool can work for them.

More types of waste

We look at all waste, but what about just recycling? Every bin in OSM can have the types of waste it accepts defined, but we aren't currently using that.

This is of particular interest to our project partners Leeds by Example and Hubbub who are looking to expand their on-street recycling bins to more places.