Project Proposal: Running Route Recommendation using Machine Learning¶
1. Introduction and Motivation¶
Running is one of the most popular forms of exercise out there. But every day, runners must decide where they want to run. Running the same old route day in and day out is exhausting, and planning out routes is usually a bit more complicated than anyone wants, so many often end up running the same route or same small set of routes. Moreover, finding enjoyable routes is especially difficult, especially for people who are new to running or new to the area that they want to run in. It can be particularly difficult to find routes for certain types of runs as well, as some prefer different terrain or different types of routes based on the workout they want from it. Data science can provide useful insights and recommendations for runners based on these common inputs in order to attempt to meet individual preferences.
2. Data and Dataset¶
The largest source of data for the project will be from geographic location data via OpenStreetMap or Google Maps. The dataset will include some of the following features, from the following sources:
- Geographic location (latitude, longitude) – OpenStreetMap or Google Maps
- Traffic volume and speed – National Highway Traffic Safety Administration
- Landmarks and points of interest – Google Maps, Yelp
- Elevation/Terrain – OpenStreetMap or Google Maps
- Crime rates – FBI's Uniform Crime Reporting Program
- Runner feedback (preferred distance, terrain, etc.)
- Type of route (loop, out and back, point to point, etc.)
- Weather patterns (temperature, humidity, wind speed, precipitation) – National Oceanic and Atmospheric Administration
3. Methodology¶
The data will be used to train a machine learning model that recommends running routes based on individual preferences and other factors that could impact the runner's experience. The model will take into account individual preferences such as preferred distance, terrain, etc., as well as other factors such as traffic patterns, crime rates, and weather patterns. The code will output an HTML file containing the route plotted on it. Here is the plain map output, with no route plotted on it:
4. Conclusion¶
In conclusion, this project aims to provide a solution to the problem of finding safe and enjoyable running routes by using machine learning to recommend personalized running routes. By providing runners with route recommendations that meet their individual preferences, we can help more people enjoy the benefits of running and reduce wasteful and often avoided planning time