Thomas Tai | Findigs+: A personalized apartment recommendation and prediction service

Applying for an apartment rental involves a lot of uncertainty in addition to a steep application fee. With Findigs+, you can know your chance of getting approved before submitting an application.
 

uc?id=1mGf_0_TucpaVqrnlcL3ankGxXOfU6aSW&export=download
First, the user will land on the homepage to select a city.

Enlarge

uc?id=1So05DKhQq_NTNpqtQsSD7_LjUw9oATl2&export=download
Next, the user will add their credit score, monthly income, and liquid assets to determine the best matches for apartments.

Enlarge

uc?id=19-2girz4ryrM3-BuJcD9pbjrRhGP_ivk&export=download
The website will return apartments ranked on the user's likelihood of getting approved.

Enlarge

uc?id=1OgPY-ZE9EBBAmFllADixJou4ISrethvI&export=download
Finally, the user can view their chance of getting approved and explore similar listings nearby.
 
 
 

 
This project aims to design and develop a new listing service for Findigs.com, a rental technology startup based out of New York City. A combination of machine learning algorithms and techniques is used to implement a state-of-the-art hybrid-based recommendation system. Using data from past user applications, the model can determine a new renter’s chance of success for a given apartment. The baseline accuracy achieved is 75% using a random forest model. This recommendation service will increase transparency for both tenants and landlords, improving the quality of applications on the platform.

 


Tags:#MachineLearning#ProductDesign#ApartmentRentals