On Line

Today the site went online yesterday. It run on GAE (Google App Engine), and the data is taken directly from Montco's Active Incident page. The data is collected everyone 5 minutes, but it's only loaded up every few hours here.

In addition to Active Alert Events, weather is also collected. On the backend, Google VM's analyze this dating, first pulling it down using python-gcloud. The data is organized using pandas, http://scikit-learn.org and https://www.tensorflow.org/. But, get quickly get things installed, we downloaded python from https://www.continuum.io/downloads.

Yes, we make use of Docker on Google's VM. I thought was to build a complete environment, then, push is out to other people interested in do data analysis - mostly township officials that are responsible for implementing policies.

As you can see, a few other services are used as well… Cloud Storage is used for storing the data for the public. Cloud Database is used to first collect the data. Cloud SQL and BigQuery are used to analyze the data.
Screenshot 2016-02-12 21.06.30


There's also an iOS app that get's live feeds. Depending on users location and preference, they'll get alerts and can be updated as to the probability of an accident happening at an intersection that they're approaching. But, note, a lot of this is just R&D at this point.

Screenshot 2016-02-12 21.10.25

A Golang server is used to broker upstream and downstream messages in the cloud.

Ironically, the goal of this project is not to make money. It's to instruct townships on the importance of monitoring traffic patterns, gas leaks, and even disease out breaks. After attending many town meetings, it's obvious to us that many people in local government could use this data. Or, at least, they would like to know how it would work — they want a working model.

References:

Sites:
https://www.kaggle.com/
Books:
"Elements of Statistical Learning", Hastie..

Also, take a look at this link from redit … here
blog comments powered by Disqus