Simple Development of Ridesharing Solver Using Reinforcement Learning (using supplied Github repo where most work is already done)- 3 days
Budget: $30 – $250 USD
I am looking for a developer to complete the development of a ridesharing solver using reinforcement learning. The majority of the work has already been done and is available on this GitHub repository:
https://github.com/UMich-ML-Group/RL-Ridesharing
Programming Language:
- The solver should be developed in Python.
Reinforcement Learning Algorithms:
- Use same algorithm in the repo
Timeline:
- The expected timeline for this project is 3 days.
Ideal Skills and Experience:
- Strong proficiency in Python programming.
- Experience in reinforcement learning algorithms.
Ensure the following:
1. Follow the Github repo intructions and setup environment with correct version of python packages
2. Minor amendment to ensure "costs" of all distances between nodes are uniform (e.g. all set to "1")
3. Run the model for some time an ensure the loss is reducing (you don't have to run it to optimality, just show the loss is reducing)
4. Save the model at some point and used the savd model to generate shared routes for a number of cars and passenger (e.g. 10 cars and 100 passengers).
***The most important thing is that the trained model (even if not optimal initially, but I can take the code and containing training) can accept a number of cars and a number of passengers and produce an *output* i.e. plan trips from passenger originating location to destintion.
https://github.com/UMich-ML-Group/RL-Ridesharing
Programming Language:
- The solver should be developed in Python.
Reinforcement Learning Algorithms:
- Use same algorithm in the repo
Timeline:
- The expected timeline for this project is 3 days.
Ideal Skills and Experience:
- Strong proficiency in Python programming.
- Experience in reinforcement learning algorithms.
Ensure the following:
1. Follow the Github repo intructions and setup environment with correct version of python packages
2. Minor amendment to ensure "costs" of all distances between nodes are uniform (e.g. all set to "1")
3. Run the model for some time an ensure the loss is reducing (you don't have to run it to optimality, just show the loss is reducing)
4. Save the model at some point and used the savd model to generate shared routes for a number of cars and passenger (e.g. 10 cars and 100 passengers).
***The most important thing is that the trained model (even if not optimal initially, but I can take the code and containing training) can accept a number of cars and a number of passengers and produce an *output* i.e. plan trips from passenger originating location to destintion.