Real Estate is an asset class like no other, where asymmetric information and traditional investment processes mean that it is perfectly poised for technology disruption. In this article I would like to explore how Real Estate professionals can leverage this technology.
Augmented Reality – Relevant Real-Time
Augmented Reality (AR) has become prevalent in gaming over the last few years. Being able to overlay a digital interactive experience onto the real world. Pokémon Go! saw the first global adoption of AR.
AR can also be applied to more practical uses. For instance, RE5Q has thousands of data sources, AR allows the user to see the data and the buildings the data relates to at the same time.
By clicking a point on the screen, the user can capture any property of interest, see more information held on that property and add that property to a list for later analysis.
Once a property is flagged as interesting, automated processes can start work: e.g AI agents find similar properties based on area, valuation, square footage, yield, or other factors important to the user. The user can be notified if similar properties come into the market. The range and scale for automation is large, covering both commercial and residential real estate.
Using AR in this manner is a unique way of capturing property information. If the user is remote, dropping a pin on a map would be another way user could add buildings.
Unlocking Unstructured Financial Data
Information is key to making both valuation and investment decisions. Being able to obtain timely, independent, accurate information can be extremely valuable. Data in the Real Estate sector is largely building specific. Building operators have an asymmetric advantage due to availability of information compared to other market participants.
A possible solution to this can be through using a set of financial reports created by REITS and property funds.
To access the information in these reports, analysts must open each document, find the relevant financial metrics, and enter them into a separate financial model or spreadsheet.
Using feature extraction and machine learning, it is possible to process financial reports, in any language, extracting the useful content such as location of assets, price paid for assets and yield obtained.
Automation dramatically reduces the time it takes to create a model, freeing analysts to focus on key questions:
- Where are other asset managers investing?
- How much capital are they deploying?
- What comparable properties are there in an area based on yield achieved?
- In each portfolio, how much is spent on maintenance?
AI can then go one step further using ‘Machine Comprehension’ techniques where the financial reports are not just found by the AI agents but also understood. The user feeds the machine an article, report or text file and can ask the machine relevant questions which it will then answer.
Visualising Big Data
Big data refers to datasets which are too large to be analysed using traditional tools such as spreadsheets. When handling this size of data, it can be hard to make visualisations and gain insights.
RE5Q holds data on almost every property in the UK. The data includes the asset level and area level information which is obtained from thousands of open and paid sources.
Visualisations such as heat maps present large complex data in a simple-to-understand format.
In this instance we are looking at commercial property, mostly office spaces and their values in certain regions of London. Without the visualisation, one could infer that central London is the only place where office space is highly valued. However, by zooming out and looking at the macro picture there is a red area around Stratford which was redeveloped as part of the 2012 regeneration programme.
Being able to visualise data this way and make an inference through images can be useful when making investment decisions. Big data enables you to see the macro picture but can also drill down to see local phenomena.
The image above, a zoom-in of Knightsbridge, shows all the office spaces located within the set radius. Using AVM (Auto Valuation Modelling) and big data, scores can be allocated to each office space based on the surrounding comparable properties, providing the users with an understanding of areas in which they are interested.
To conclude, the examples discussed in this article demonstrate a small slice of possibility when applying AI to Real Estate. As data improves over time, these possibilities expand and with the scalable technology outlined above, we will be able to grow with it.