Signal 5: To Rehab or Not To Rehab?

Rapid growth, followed by a period of suburbanization and instability, ending with a gradual rebirth of the core: that’s the last hundred years in a nutshell for most American cities.

The City of Memphis faces the problem of rehabilitation. It’s hard to detect all the community that need restoration. Assessing every single community to decide whether it is blight is costly. However, the data science can do it at ease.

Using the random forest classifier, combine with administrative data, to identify properties as “distressed” or “not distressed”. A more nuanced estimate of a property’s condition than a simple binary distinction.

Turned these results into a web application that displays a color-coded risk assessment for every residential property in Memphis. Notice that the highest-risk areas correspond to the blue parts of the classifier map, as you might expect.

Furthermore, we can match each house in a neighbourhood with the five recent sales most similar to it in size, age, and quality. Then adjust the sales price of the comparables using coefficients from a hedonic regression of house prices, and averaged them to get an estimate of the total assessed value. By varying the inclusion or exclusion and sale price of a specific property, we can see what the impact of a hypothetical renovation might be.

This can be the basis for another web application that allows policymakers to explore the potential impacts of renovating various houses in the neighbourhood.

Reference:https://dssg.uchicago.edu/2014/11/12/easing-the-distress-of-neighborhoods-with-data/