gregfrasco / real-estate-price-predictions Goto Github PK

Create a card component to feature a listing.

Takes in listing as a prop.Wireframe
Material Example: Look at the image card. Example

Example Source Code

List the number of Bedrooms, Bathroom Sqft, and Garage if applicable.
Look at using Material UI's Chip Component for this.

Wireframe

If we choose to use LOTSIZE in our models, we need to clean the data.

• Some lot sizes are listed in acres, and some in square feet, skewing the data significantly
• Some lot size is 0 or extremely high, which should be validated

For this first pass, we ended up dropping some of the non-numeric features of the data, like style, dates (used days on market, but not the list and sold dates), cooling/heating, or other info like basement, fireplace, etc.

Not all of these will be relevant, but some might be. If anything stands out to you, you can play around with one-hot encoding additional features: https://towardsdatascience.com/categorical-encoding-using-label-encoding-and-one-hot-encoder-911ef77fb5bd. They will be easy to add to the model later on.

Get flip score given a MLSnum

Given city - get listings by city

According to a realtor friend, one tell-tale sign of a fixer upper is a small number of photos attached to the listing. To get this in, we don't need to analyze the photos themselves, just add a column to the data that says how many photos were included in the listing.

The imgs.zip file that we received has images with the MLS number as part of the filename. I'm not sure it has every listing, but we should be able to count what's there and add it to the data.

note: We'll need some way to treat listings with 0 photos in a special way, since that means we don't have the photos in the zip, not that there were no photos included. We don't want this to skew the data. But let's try and figure out what we have.

Using Material UI Select Component

Create Map Component

Some values for age are showing up as a year, ex "2010", which is really messing up the data. This should be cleaned up to be useful!

Create index page

Card a card with the list of

• Estimated Sold Price
• Estimated Price After Flip
• Profit
• Flip Probability

Wireframe

Convert Search into a select component. Values will be city values from our dataset.
https://material-ui.com/components/selects/

Take Listing in as a prop

Render on a single card

• Home image
• Address
• City
• Listing Price

Wireframe

Create a DB Model for listings

Add Listing Description Component

Wireframe

Get All homes in a select city

The following listings need to be cleaned. Here is the process:

Paste MLS # into Redfin search
Find listing the the correct csv (see sales date)
Fix listing details
Save. Commit and push changes.

72250832 - SOLD Jan 2018
71902243 - SOLD Jan 2017
72214658 - SOLD Oct 2017
72099376 - SOLD April 2017
72032454 - this looks to be a listing in san diego, so the MLS is probably wrong too
72027853 - SOLD Nov 2016
72018311 - SOLD May 2017
71955378 - SOLD Oct 2016
72045937 - SOLD Dec 2016
72133139 - SOLD May 2017
72144618 - SOLD May 2017