This page contains the results of this project. The other parts are:

Results

Once the Machine Learning model built previously has been trained, it was used to predict the rent for sale and auction listings, and the results were used to create a Tableau dashboard. This dashboard is thought for potential real estate investor in Italy and it assumes that a property can also be bought through a mortgage, in which case it allows the user to edit the details of the loan like the interest rate, the down payment and the term. It also allows the user to include into the cost of an investment the “renovation relative cost”, i.e. the cost of renovations as a percentage of the purchase price.

The dashboard showcases two different investment metrics:

annual cash-on-cash return: annual net income (rental income minus mortgage monthly payment) divided by the down payment
rental yield: annual rental income divided by the purchase price

These can both be useful in making investment decisions.

Here is the dashboard:

The dasboard can also be opened in a new window here, or it can be viewed in full screen mode using the control in the control bar. It is rendered best on large screens (not smartphones).

Here is a short (non comprehensive) list of examples of useful insights that can be obtained from this dashboard:

There is no clear geographical trend for profitability when looking at the whole dataset: there are profitable areas both in northern and southern Italy
Listings on auction are much more profitable than listings on sales (the reason is simple: the same property will be cheaper if it’s being sold on auction then on sale)
Less energy efficient properties are generally more profitable, even when accounting for significant renovation costs (the reason is again simple: all other features being equal, a property will be much less expensive if it’s in a much less energy efficient class)
Rural properties tend to be much more profitable than others, even when significant renovation costs are taken into account