/DS5110Project

This project is part of DS-5110 course Data Management and Data Preprocessing

Primary LanguageR

A very useful website

https://rstudio.github.io/shinydashboard/structure.html#background-shiny-and-html

Using Docker

You don't have to use Docker. But if the app is giving you installation problems, this is a good approach.

Building and running a Docker Container

  1. Install docker
  2. Build the docker container: $ docker build -t boston .
  3. Run the docker container: $ docker run --rm -p 3838:80 boston
  4. You should see the app at localhost:3838
  5. Get your the process number of the docker container: docker ps
  6. Open a different terminal & connect to the Docker container: docker exec -it <mycontainer_process_id> bash

Editing code with Shiny Server and Docker

  1. Connect to Docker container
  2. The code is located in /srv/shiny-server/
  3. Edit code with vim, emacs, or nano.
  4. Save the app.R file and referesh, changes should flow through
  5. Clone this repo somewhere in the docker container and copy your changes to the appropriate place in this repo before committing.
  6. Make sure to reference the issue number in your commit (see #7).

DS5110Project

Title:

Property Assessment Visualization for the City of Boston

Authors:

Tyler Brown , Sicheng Hao, Nischal Mahaveer Chand, Sumedh Sankhe

Summary:

When we put ourselves in the shoes of potential Greater Boston area homebuyers, we find they have many informational resources available to them. As a homebuyer, their major sources of information are the websites Zillow \cite{ZillowRe12:online} and Trulia \cite{TruliaRe98:online}. These resources provide information such as home location, price, amenties, and types. Trulia differentiates itself by providing a ``Local Scoop'', giving the homebuyer maps of crime, school location, and the relative distribution of home prices in Greater Boston. However, neither of these websites provide information about how neighborhoods are changing over time.

Homebuyers want to better understand their potentially new communities. It's helpful to know if your neighbors have been regularly developing and maintaining their properties. A homebuyer would also like to know whether a neighborhood has been experiencing various levels of turnover. Our group is focusing on the problem new Greater Boston area home buyers face when they want an intuitive way to understand changes within their potential neighborhoods over time.

The City of Boston provides an open data platform, Analyze Boston, containing information related to our lives in the city. Property assessment data from 2014-2017 is one of the resources available on their open data platform. Included in the property assessment data is ``property, or parcel, ownership together with information about value, which ensures fair assessment of Boston taxable and non-taxable property of all types and classifications'' \cite{Property49:online}. Our team has aggregated each available year to create a time series dataset of Property Assessments in Boston. This aggregated dataset allows us to provide unique insights into property valuations and ownership strategies.

Proposed plan of research:

The dataset we have right now is separated according to year in different files. We will start by merging the data into a single file and completing the necessary cleaning and data wrangling steps. We will then build a web application which provides the user with a helpful dashboard. This dashboard will include a selection bar for their home preferences, and a way to select which changes in the neighboorhood they want to explore such as assessment changes or remodel status. We plan to improve exploration of neighborhood changes by applying a clustering model to see if patterns exists. The dashboard will also include an interactive map to help users visualize those changes.

Preliminary results:

See "proposal.pdf" for preliminary results.

References:

Outline:

Part one:selection bar(sidebar of properity information selection)

Part two:tabs(models)

(After select properity type) Model_1: Remodeled properities and prediction

Visualize TR_BUILD and YR_REMOD within each zipcode

Predict how much properties are likely to be remodeled (by each zip code region)

and then visualize density

Model_2: Interior detail

Bath style, kitchen style, heat, AC, interior finish, interior condition, view

Visualize each part of each zipcode

Model_3:Assessment value change(need wide format)

The year 2014-2017, find the key to merge our data into a wide format DataFrameCallback

find assessment value change(not necessarily everything)

visualize the change and prediction of the future years(two or three years)

Part three: map