Some of the latest advancement in the field of Machine Learning (ML) has really opened doors to solve some long-lived problems in a very elegant way and Real estate vertical is not an exception. The more I see its application and outcome, the more I am convinced to directionally use it for long lived elegant solutions to some of the harder problems of this sector.

Machine learning is a sub area of machine learning, which gives computers the ability to learn with­out explicit programming. The traditional way of solv­ing a problem has been to codify a rule based system based on expert inputs. But this only works properly when experts have complete knowledge of all the rules of the ecosystem. 

When the complexity of the system is beyond a point, then orchestration of which signal is influencing an outcome in what way is beyond control. In such complex systems these rule based tra­ditional expert driven systems start to fail. In such an evolving and complex situation, use of machine learning techniques have proved to be much more effective.

For real estate platforms, ML techniques can be used for various important use cases.

Benefits of us­ing ML techniques:

1. Automated valuation ma­chines/ Price estimators:

The price of a house is influenced by various factors like location, brand, size, amenities, construction quality, age, facing, interiors etc. To have a rule based approach to determine which factor is influencing price how much is almost impossible beyond a point, and in such cases, machine learn­ing based approach scores really high. If you train your models with huge number of listings data with price, the system will automatically figure out what is best for a given input data.

2. Fraud content detection in listings:

In a country like India, where there is no central source of listings (MLS), it is very important to have a quality control on the list­ings quality to deliver a higher qual­ity experience. To verify millions of listings on the platform manually is non-scalable. The machine learning approach has really proved to be a boon in such a case wherein we have trained millions of listing data (im­age and other content) and reached a level where the identification of real estate images, their classification started happening automatically. Even on the non-image listing con­tent the NLP along with ML works really well.

3. Lead scoring:

This is a classic problem across verticals. The good thing about real estate is if the jour­ney is longer, the user provides many signals before transacting a proper­ty, which enables us to the lead scor­ing in an effective way. If one com­bines the click stream data on the website along with CRM data col­lected at various states of journey in sales, then a very effective lead scor­ing system can be developed which can be used to effectively utilize the sales productivity.

4. Sentiment analysis on user gen­erated content:

Real estate is such a big decision that before going for transaction users discuss and measure the decisions on various parameters. In case you have a platform, which has a lot of UGC, then you can use machine learning to come up with the sentiment analysis of a given de­velopment, neighbourhood, develop­er, etc. This can be further leveraged for better decision making and repu­tation management.

There are many such applications of ML in real estate e.g. Under­standing the context of conversation from a recorded audio conversation, Automatic response by bots on real estate queries based on the knowledge base across platforms etc. So far, we have just explored the obvious use cas­es, but the possibilities are immense.