Data Challenges

DataHub Ruhr is a business building program that connects start-ups with established corporations in the Ruhr region. The program tasks start-ups with developing innovative, data-driven ideas to tackle use cases provided by our corporate partners. As part of a three-month collaboration, start-ups will draft a proof of concept, with the opportunity to receive up to EUR 20,000 in funding.

No items found.

Tenant Happiness :)

As one of the leading housing providers in North Rhine-Westphalia, VIVAWEST manages more than 120,000 apartments on the Rhine and Ruhr rivers and around 300,000 people find a home here. As part of a sustainable business model, VIVAWEST combines economic efficiency with ecological and social responsibility for customers, employees and the region.

Website

Startup

Result

Keeping reliable tenants happy is of high importance to landlords. There are many early indicators before a tenant cancels their contract. Their decision might be explained by increasing rent or by complaining about the same issue several times. Can we develop an early warning system for tenants that are about to leave?

Use Case

VIVAWEST owns more than 120,000 apartments in the province of North Rhine-Westphalia. Roughly 8% of tenants cancel their contracts every year. Reducing that number just a little would have a tremendous monetary impact, but could also substantially improve the satisfaction of thousands of tenants.

Our goal is to spot tenants who are about to cancel their contracts. In a next step, these tenants could then be contacted to see if a resolution to their problem can be found. A variety of sources influences tenants’ decision, e.g. changes in the neighbourhood, rent increases, damages in the house/apartment or mistakes in the statement of utility costs. Additionally, tenants might be sending signals before they give up and cancel, e.g. they might inquire about the cancelation policy or complain about problems.

This data has not been analysed yet since different data sources need to be combined and the influence of each factor is uncertain. As of now this is done by experts and their individual experience. We would like to develop a system that gives a prognosis of how likely a tenant is to leave within the next six months.

A project duration of approximately 2 months is expected.

‍

What you will need

Experience in the field of tenant management or real estate
In-depth knowledge in (churn) prediction, survival modeling and text mining
Proficiency with survival models (proportional hazard etc.), logistic regression or other classification models
Ability to handle large amounts of (semi-) structured data from different sources

‍

Expected result

Reliable forecasts: cancellation probability within six months
Very high true positive rate/sensitivity: we do not want to contact people who didn’t plan on cancelling
Any noteworthy relation between data points and/or external influence factors shall be documented

‍

MILESTONES

The project can be divided into three milestones:

The first milestone is reached when the available data has been cleaned and brought into the right format for further analysis. Visualizations and descriptive statistics have been used to understand the problem of identifying relevant tenants.
The second milestone is reached when a model/machine has been developed that accurately predicts which tenants are about to cancel their contracts.
The third milestone is reached when the algorithm makes suggestions about which customers to contact right now and when contacting these tenants confirms the prediction. The result is presented to and accepted by VIVAWEST.

‍