We all like to live in a clean city. Therefore it is nessesarry, that dirty places will get cleaned with the shortest way to drive. That will save time, money and most important carbon dioxid.
What it does
- We built a Heatmap showing off the contrast of dirty to clean places in the City of Basel. The recipient for this service is a non-technical user.
- We further optimize cleaning routes via calculation of a distance matrix. The optimization goal is to reduce the carbon footprint via reduction of vehicle distances whilst taking also into account, the clean-index of an area.
- We predict dirty places going one day into the future. The algorithm uses the day of the week, special events like vacations and the weather to filter relevant data and put them in the proper context.
How we built it
The prediction is build in Java. The frontend is made in Python as well as the distance calculations and the route optimizations.
Challenges we ran into
Often Timestamps and data were corrupted. Also we had only 250 messures of every day. Also it was hard to fit the new Data into our own build backend. It is hard to find a way for a cleaning car with the background data of a Bike ;). Using the right distance metric is always challenging and cannot be considered optimal. Further data from the city of Basel might be a big improvement.
Accomplishments that we're proud of
The team, the challance, our outcome and many things more.
What we learned
To have a very good forecast you really need a lot of data. The easy features are not always easy to addapt for our needs. It is all about the data.
What's next for Haxxeta cleaner city
We would like to change the base of the data (Pictures from cleaning cars) so with that we are preatty sure we would be able to crate nearly optimal routes for the cleaning cars of tomorrow.