The world’s leading publication for data science, AI, and ML professionals.

Data Science for Real

Transforming property management with advanced analytics and machine learning models

_"It’s tangible, it’s solid, it’s beautiful. It’s artistic, from my standpoint, and I just love real estate." -_Donald Trump

I usually don’t agree with President Trump. In fact, quite the contrary. Although – concerning real estate – it seems we have a shared passion.

Property has been synonymous with wealth for thousands of years, has helped amass great fortunes and will probably continue to do so for generations to come. One of the key enablers for generating value through real estate investments is through sound property management.

While time-tested techniques have been developed for managing property, many of those methods were developed in an analogue world. We now live in a time where devices are getting smarter, everything is connected, and algorithms are taking our jobs. We have moved from the analogue to the digital. The property manager still has an important role to play in this new digital age, but we are now increasingly able to make him more efficient and let him spend time on the tasks that have a greater business impact.

There are several ways real estate companies can leverage Data Science to improve property management – from integrating smart building technology to implementing machine learning models for tenant management. This article will discuss some of the ways to prepare for the use of these technologies, and present concrete use cases for advanced analytics and machine learning models within property management.


Walk Before You Run

Using machine learning algorithms to drive profits sounds cool – and it can be a great driver of value, but there needs to be an understanding of business processes, structure and governance in place before these technologies can be fully utilized.

Process

Firstly, the business processes should be clearly understood and defined. Business process management, or BPM as it is referred to by practitioners, might have fallen out of fashion, but there can still be valuable insights gained from mapping out your processes and analyzing them. BPM can give a clear understanding of the current processes and makes it easier to see what parts of a processes can be improved with new technologies and which should be kept as is. Take a modular approach to your processes and analyze the inputs and outputs required in each step. Always consider ways to simplify and standardize, as this will make digitization and automation much easier.

From Spreadsheets to Databases

Spreadsheet programs like Microsoft’s Excel have had great success. They are used by millions of businesses around the world, and for many ad hoc tasks they work great. But when it comes to reporting and more advanced tasks, spreadsheets can be cumbersome and prone to errors, as well as not being able to provide the right data you need for your models. Sure, you can make your sheets smart by integrating Visual Basic code and create great macros, but this can be difficult to develop a potential nightmare to maintain.

By its very nature, being easy to manipulate and change, the spreadsheet is not an ideal solution for reporting and other routine tasks. It will almost surely lead to an increase in errors and ambiguity. Moving away from spreadsheets and towards a more database-like world, makes it possible for machine learning algorithms to be applied to problems much more efficiently. Tasks can be analyzed and automated to a much greater degree in a more structured system. This is critical for companies that want to take the step into data science.

Data Governance

How you handle your data, how the data is secured and stored, who is allowed to insert, update or delete a record are key questions within data governance. Enterprises benefit from data governance because it helps to ensure that the data is consistent and trustworthy, as well as in compliance with regulatory constraints.

From a data science perspective, the quality of and consistency of the data is of utmost importance. The old saying, ‘garbage in, garbage out’ is especially relevant when training and using machine learning models. Without access to good data the models will be weaker and their output and predictions will be less trustworthy. It is therefore strongly recommended to have a good data governance process in place before implementing advanced machine learning projects.


Using Data Science

With the right processes, data governance and architecture to support it, it is time to venture further along the Digital journey. Data science – the application of advanced analytics and machine learning models to industry problems – is the next natural step. Below are a few examples of how this technology can be used to enhance property management.

Tenant Churn

Churn modelling is one of the classical applications of data science. Industries such as banking, insurance and telco have been using churn models for decades to predict customer behavior, and it has obvious use cases for property management as well.

In its simplest form a churn model is a binary classification model that given a set of predictors, or input variables, outputs a classification. For instance, if a tenant rents a unit, a churn model can be used to predict the probability that tenant will leave within a given time frame, for example 1 year.

Armed with this knowledge a property manager can better understand how the tenant mix will change over time and which units are likely to become available soon. This will enable managers to proactively target the high churn candidates and provide incentives for them to extend their contract. In addition, the model can highlight which units likely will need to be filled in the future, so that the vacancy period of a unit can reduced.

Churn models can also be used efficiently as a component of larger forecasting models, where the probability of churn is factored into future cash flow streams and aggregated across a broad population of tenants. By doing this, a real estate company can make more accurate predictions of its cash flow and therefor increase its total leverage without a proportional increase in risk. Ultimately, leading to a more efficient use of capital.

An example of a real estate company who takes tenant management seriously is Spire Property Management. They use MRI – a dedicated real estate software solution – to assist in their churn prevention, among other things. According to Sean Paul, and executive director of Spire they use MRI to "streamline customer relationship management… organize, automate, and evaluate its retention efforts – keeping track of leases and lease anniversaries, tenant records, maintenance activity and work orders". Effectively creating win-win solutions for both property managers and tenants.

Lead Generation For New Tenants

Do you know who your best tenants are? Are they the ones who have the highest willingness and ability to pay, or perhaps the ones who always pay late so you can collect late fees? Or perhaps the best tenant is the one with a short one- to two-year turnover, giving you the opportunity to create new contracts and increase the rent. The answer to this question cannot be answered by a Machine Learning model, but should be evaluated from a business standpoint, aligned with the companies’ overall strategy.

Once the optimal tenant has been clearly defined and a company knows who it wants as tenants – keep in mind this can vary between buildings – it is possible to use machine learning algorithms to map out the features that are characteristic of these ideal tenants. This feature set can then be run through various leads-lists to determine which have the potential to be the best tenants. When the feature set is rich and complex, machine learning models often perform this type of classification at superhuman level, and scale infinitely better than humans.

For commercial real estate there are also several providers that help companies generate leads and increase sales. One example of such a company is Vainu. Founded in 2013, they are using machine learning models paired with large databases to assist in the sales process. A case study from the real estate company Technopolis, revealed that they were able to use 3 times less time on sales prospecting than before. This type of solution could be especially relevant for property managers looking to increase their lead generation for their commercial units.

Location Optimization for Employees

Consider the following scenario. A company owns two office buildings with several units and have employees that use both buildings. Due to unexpected circumstances there is an imbalance in the usage of office space, which leaves one building over capacity while the other is under-utilized. How can this sub-optimal situation be improved with technology?

Using analytics and sensor systems the location of employees can be monitored, and an automatic messaging system can be configured to inform employees of the imbalance and guide a subset of the group to the less occupied space. This type of optimization exercise is not difficult to implement but can provide substantial savings if imbalances frequently arise, and can lead to less overall usage of office space.

Smart Building Technology

Smart building technology holds the promise to greatly benefit property managers. Some of the innovations are rather new and their potential not fully explored yet, but there are a few use cases that really stand out.

Predictive Maintenance

Predictive maintenance can help you identify issues with buildings before they occur and thus take action to prevent a breakdown or limit downtime as a result of repairs. This can make the maintenance staff more effective and thus reduce costs, as well as potentially increasing the total lifetime of appliances. For tenants as well, having a shorter downtime on critical infrastructure repairs will be advantageous.

The first step in using predictive maintenance techniques is to add sensors to the equipment that needs to be monitored. Data from these sensors is then collected in a database, and after some time in operation, time series data about the operation of the system is accumulated. Given that we have enough observations, the time series then contains the data needed to build machine learning models that predict the next failure of a system.

These machine learning models are typically either classification models or a regression models. In the case of a classification model we will try to predict what the probability of failure will be within the next n time steps. With a regression model we predict how much time is left before the next failure, and this is often referred to as ‘remaining useful life’. Whichever model is chosen will depend the type of system and available data for modelling.

Building Information Modeling

Building Information Modeling, or BIM for short, is another exciting technological development within the smart building space. The basic idea is that the building has an identical digital twin complete with schematics and 3D-models. While many construction companies use BIM to aid in their building process, they can also provide significant advantages over the lifetime of the property, and thereby assisting the property manager.

A typical example of the usefulness of a BIM for Property Management would be when a system in a building failed. Since the building has a complete 3D-model of itself, the maintenance engineers can use augmented reality while traversing the building and obtain critical building information while performing repairs. Data that could be provided to the engineers include service history, system specifications and contract information, making the repair process easier and faster, and as with predictive maintenance also lead to shorter down times for critical infrastructure.

Beyond Property Management

It is important to note that we have primarily discussed how digitization and data science can be used to improve property management. This is of course just a subset of the Real Estate industry and there are a plethora of other possible application areas of data science for the industry as a whole. Machine learning models are now used to predict anything from price and rent income to demographic trends. As we continue to use more IoT devices and algorithms get better we will surely see many more use cases develop.


There are many ways property management can be enhanced by the use of emerging digital technologies. But to take full advantage of the recent advancements in analytics and machine learning, an organization requires a certain degree of digital maturity. However, once that threshold is reached, data science can be used in a variety of ways. Savings can be generated from having a better understanding of tenants and their churn or by using predictive maintenance techniques and BIM models, and sales can be increased with lead generation models.


If you enjoyed this article and would like to see more content from me, or would like to engage my services, feel free to connect with me on LinkedIn at https://www.linkedin.com/in/hans-christian-ekne-1760a259/ or visit my webpage at https://ekneconsulting.com/ to see some of the services I provide. For any other questions or comments please send me a mail at [email protected] .

Thanks for reading!

References:

MRI Software – Liberate your real estate business

How tenant churn could impact profitability

Machine Learning Techniques for Predictive Maintenance

What is data governance (DG)? – Definition from WhatIs.com

What Is BIM | Building Information Modeling | Autodesk


Related Articles