A Data Scientist goes…House Hunting (Part 4 – Searching for Value)

Note: The finished dashboard for this project can be seen here.

So far we’ve built up some basic understanding of the overall market and made some visualisations that allow us to peruse the properties for sale by geo-location. That’s all good but not much, if anything, more than what you can do on most of the listings sites. What we want to do now is start developing insights that you can’t see on those sites with the aim of identifying value vs price.

Some ideas that spring to mind are:

• What’s the motivation of the seller?  Are they aggressively reducing price or has the listing sat just on the market for months and the seller is waiting for the market to come to them?
• Are there properties that appear to be having difficulties completing the sale? Sometimes properties go Under Offer or Sold only to come back onto the market a few weeks later. Such situations might offer opportunities in value depending on what the reasons were.
• How is the property priced vs its peers? One could perhaps compare to similar sized homes in the same area that are on the market (the competition) or have sold recently (indicating the true value).

To develop these insights we need to analyse the listings histories and develop some more features for our model.

Once again, we will be using Jupyter Notebook to develop our code: