2019.3.29 - 2019.3.31
Python, Google Cloud, Google Map API, Flutter.io, Figma
Won the Sponsor's Honor
Competition: LA Hacks 2019
Team Formation: 4 software engineers and 1 UX designer
Goal: To create an app in 36 hours which helps LA citizens find the safest route among all routes Google Map provided
According to a 2014 Crime Survey, 37% of adults feel unsafe walking alone at night in the US. Los Angeles, in particular, has a crime rate higher than 84% of the state's cities and towns of all sizes. Keeping this in mind, we believe it is crucial for residents and visitors alike to be aware of the safest streets to take at any time. We decided to build an app in order to help passengers choose the least risky path they should take, focusing on the LA area.
"Safe Route" is the dream app for pedestrians who wish to navigate their daily life safely and efficiently! Once the passenger inputs a destination on Google Map, the application automatically calculates its own safety scores (-2 to 2) for each route, based on historical data on crime rates. The user is then allowed to preview the routes and decide which path to take in Google Map.
Furthermore, "SafeRoute" is time sensitive, and therefore provides an up-to-date and accurate estimation for every two hours of a day.
Machine Learning model using the LA Crime Data Set
How We Built It
"Safe Route" was built using Google's new Flutter language as front-end, and Machine Learning algorithm as back-end. Specifically, we coupled the publicly released LA crime datasets with the robust Google Maps APIs to enable us to provide an insightful safety score. This score was determined by calculating the relative frequency of crime in each area and surrounding areas.
We want to make sure everyone could look at our safety score to get a quick understanding of what to expect for a given path, so we incorporate color, meaningful content and visual representation on the UI design. Furthermore, in order to ensure the safety score is accurate, we reorder our data set according to the time input and run a separate algorithm every 2 hours.