Problem

Our team was tasked with creating a reliable SBB journey planner for the Zurich area using data from the Open Data Platform Mobility Switzerland.

We encountered several obstacles during the project, including incomplete or inaccurate data, limited computing resources, and time constraints. Additionally, we had to ensure that our solution was user-friendly and accessible to a wide range of users, and incorporates the delay prediction which is a very important feature.

Solution

To address these challenges, we developed a web-based application using PySpark, the Python API for apache Spark which is needed due to its distributed environment. We used various libraries such as Pandas, NumPy, and NetworkX to preprocess and analyze the data. We also implemented an algorithm that could handle multiple user inputs simultaneously. We made sure to use historical data to understand and quantify delay for each suggested route, depending on its time and bus/train lines.

The application allowed users to input their starting point and destination, select their preferred mode of transportation (train or bus), and view all possible routes with estimated travel times, and estimated probability of success. This enables the user to chose how much risk is he willing to take in order to arrive at chosen time.

Impact

Increased efficiency for commuters by allowing them to compare different transportation options based on estimated travel times.
Suggest some risky options for the users, that can help him gain time or effort, having an estimated risk value on each suggested option.

"Being able to chose between different options based on risk is a new feature in Journey Planner solutions. "

Team

Ghassen Karray, Seifallah Hanafi, Ahmed Ben Haj Yahia - Data Scientists
- Preprocessed and analyzed data using Pandas, NumPy, and NetworkX.
- Implemented the optimal path search algorithm.
- Implemented an algorithm to handle multiple user inputs simultaneously.

Zurich SBB Journey Planner