SaferWalk

• Built a website capable of recommending safer routes to pedestrians based on data provided by the Police.
• Deployed the graph-based algorithm on a Google Virtual Machine using Python, Flask, OpenStreetMap and NetworkX to allow for a more flexible architecture design.
• Analysed the data using K-means clustering and modelled a bivariate multimodal probability distribution using SciPy.
• Reduced the Flask API response time by a factor of 4 by approximating the heuristic function of the A* algorithm using Riemann sums and by pre-processing lattice points values