Christian Medeiros Adriano edited this page Jul 8, 2016 · 19 revisions

Context

The first version of the tool was developed during a data science hackathon that happened from May 14 to May 15, 2016. The data files used were provided by the hackathon organizers and comprised approximately 30 GB of data in CSV format.

Hackathon team

  • Christian Adriano (Graduate student in Software Engineering)
  • Chris Gala (Undergraduate student in Computer Science)
  • Sreeja Sthothra Bhashyam (Undergraduate student in Computer Science)
  • Siddhartha Desai (Undergraduate student in Computer Science)
  • Thomas Kwak (Graduate student in Software Engineering)

What it does

Our software allows companies to discover apps that are the most profitable to advertise on. We display a map that displays the users' hourly mobility, whether they're at home, commuting, or at work. We then show a tree map that visually represents a categorization of the apps based on their mobility.

Classification rules

  • If the coordinates of different data points from the same device are not within a quarter of a mile radius, then categorize the device user as "commuting", otherwise, we apply another two rules
  • If device user is not commuting and the timestamp of the data point is within business hours (from 8am to 5pm), we assume the device user is at work
  • If device user is not commuting and the timestamp of the data point falls outside business hours (from 8am to 5pm), we assume the device user is at home

Overall architecture

  • Database: stores the parsed and categorized data about mobile devices (latitude, longitude, timestamp, apps, mobility category)
  • File parsing and loading: infra-structure to read, write, process, and test large data files
  • Classifier: applies categorization rules for the mobility of the device user (home, at work, or commuting)
  • Merger: joins the information about device location with the apps the user is potentially utilizing within an hour
  • Controllers: query the database and format the data in JSON format to be visualize on a Google Map and on a D3.js TreeMap
  • Web page: provides GUI to request data by date and presents the results on a Map and a TreeMap (check the Screenshot at the end of this page)

Technology

###APIs adopted:

  • Google Maps API
  • D3.js TreeMap

Languages adopted:

Javascript, PHP, Ajax, MySql, and Java

###Data exploration We also used Tableau to perform some data exploration. Tableau charts

Final tool!

Sample screenshot: Screenshot of tool

Clone this wiki locally
You can’t perform that action at this time.
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session.
Press h to open a hovercard with more details.