Cycle Safe is a navigation system that recommends safer cycling routes by evaluating the risk of routes suggested by Google Maps.

Systems:

• Frontend: Google Maps and very simple JavaScript
  • Doc: User Application (Frontend)
  • Src: https://github.com/YoinkBird/cyclesafe_server
• Backend: Python REST-API
  • Doc: Architecture
  • Src: https://github.com/YoinkBird/cyclesafe_server
• Machine Learning Framework: Classifier Model Framework based on python, pandas, scikit-learn, etc
  • Doc: Machine Learning Framework for Model Lifecycle Management based on CRISP-DM
  • Src: modelmanager

Model:

• Risk Assessment Model:
  • Design: Predictions based on Route Data

Model generation, system documentation, and assorted helper utilities for the Probabilistic Routing-Based Injury Avoidance Navigation Framework for Pedalcyclists Project.

This is a rudimentary custom machine learning framework using scikit-learn and pandas.

This code (mostly automatically) manages model generation and validation/evaluation (train+test on TxDoT Crash Data), as well as scoring navigation routes (based on Geo-JSON).

Modules:

Start with the cyclesafe server project, which will manage all dependencies.

This repo corresponds to the Scoring Application and Modeling Application in the Architecture overview documentation.

See the Contribution Guidlines .

The docs directory contains:

• articles: Developer writeups on interesting decisions and operations.
• system documentation : Overall systems documentation, technically originated as a masters report, hence the name.