aaronnech / Zillow2015

Zillow 2015 Hackathon Project

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Description

This is the CasualHome Project from two humble UW CSE Developers Aaron Nech(@aaronnech) and Daniel Noteboom(@djnoteboom) for the Zillow Winter 2015 Hackathon.

We believe that searching for a home should be relaxing, not stressful! CasualHome is a mobile application where we seek to help you by learning as much about you as possible and then build a profile to present homes to meet your needs. Are you low income? No problem! we'll slowly provide you information about various government programs and refer to people and friends that can help you. While using the app, a personalized profile will be learned over use (swiping homes, taking quizzes, and filling out demographic information) to help us find the right home for you.

CasualHome utilizes an aggregated data model and profile intelligence to only show you houses you're interested in from many different data sources. We scrape the web from sites such as Craigslist and take advantage of many external datasets such as Accessible Homes Data(is grandma visiting?) and crime rate data(do your kids want to play outside?) to find the perfect home for your needs(by matching home profiles with your personal profil). What goes into the profile? Many different features such as your weighted preferences for commute-time, great education for your children, or the number of bedrooms. Sound stressful? Don't worry! Simply use our app, and we'll do all this behind the scenes without you having to worry about it.

More curious about how this project works behind the scenes? Check out our diagrams to see how we collect and process the data, and dig into the code yourself. Maybe you'll find a useful component that you can use in your own application.

CasualHome

Technologies, APIs, and Datasets Utilized

Technologies

Client

  • TypeScript
  • React JS
  • Browserify
  • JQuery
  • HammerJS
  • Cordova (PhoneGap)

Server

  • TypeScript
  • ExpressJS
  • MongoDB

Datasets

  • Seattle Crime Rates (data.seattle.gov)
  • King County Accessible Homes Data (src/server/data/source/kingcountyaccessibility.json)
  • Craigslist (http://seattle.craigslist.org)

APIS

Seattle Data API (data.seattle.gov)

Architecture

Client Overview

Image of Client

Server Overview

Image of Server

Harvester / Data Pipeline (detailed)

Image of Harvester

Set Up

Run the following from the root directory:

npm install

to install project dependencies into a folder called node_modules

And then:

npm run-script serve

To compile all TypeScript, and launch the main.js server located in the server directory, you can instead run:

npm run-script make

To just compile all TypeScript.

Methodology Overview

Both Client and Server are written in TypeScript and ultimately compiled to JavaScript. This is cool because:

  1. Both Client and Server can share code. For example the client and server both use a Pizza object for various functionality, we only need to write that object once (this kind of code I'm calling common and goes in the src/common directory).
  2. We use NodeJS to run server code on a server computer
  3. We use Browserify to scoop up all the seperate JavaScript files and bundle them into one JavaScript include for the browser platform.

Directory Breakdown

There are three main directories in the source:

  1. src/server - This contains all code only pertaining to the server side application.
  2. src/client - This contains all code only pertaining to the client side application. Browserify is used with the start point client/main.ts to scoop up all client code and bundle it into client/static/js/main.js
  3. src/common - This contains all code that is shared by both client and server. Including files such as typescript definitions. NodeJS processes dependencies on its own via support for require(), so any src/server code that depends on common code will do so automatically. src/client code will scoop up common code via the same mechanism, but employed by Browserify.

Vendors

Vendor files are files that do not naturally belong to the application, but need to be used (e.g. libraries). If a library has an equivalent NPM package, you can just use require() as normal after installation via NPM and both the client and server will have support to get the dependency (the client via Browserify).

If the library is client and does not have a NPM package, you will need to place it in src/client/static/js/vendor where it can be safely allowed by the gitignore. You then need to include it manually through the index.html file. This is rarely ever necessary though because NPM has literally everything.

PhoneGap Android Support

To run the application on android simply run:

npm run-script android

Assuming you have all the dependencies installed (e.g. cordova, android SDK, emulator / deviced plugged in). This script will create a _cordova directory which will contain a copy of your post-compiled src/client/static directory, and some settings copied in.

Development Style

The application development style I follow is similar to many Java applications:

  • Every class resides in their own TypeScript file
  • Every class TypeScript file ends with an export = ClassName; statement to make itself available for import
  • Every class that depends on another class, imports that class via import ClassName = require('/path-to-class/ClassName');
  • Every class TypeScript file is named as ClassName.ts

License

MIT (see LICENSE.md).

About

Zillow 2015 Hackathon Project

License:MIT License


Languages

Language:JavaScript 48.4%Language:TypeScript 43.2%Language:CSS 7.8%Language:HTML 0.6%