Enhancing Performance with Data Science

The data science age is well upon us — except, it seems, in the world of real estate.

While quantitative, systematic investing has become an entrenched norm in public markets, less than 20 percent of commercial real estate (CRE) investors use data science to support investment decision-making. This is a tremendous loss of opportunity for real estate investors, as data science offers significant informational and analytical advantages that complement traditional modeling.

To help CRE harness the benefits of data science in real estate, Climate Alpha has put together a short guide on how data science can be used to enhance market analysis. We also explain how our data lake, tools, and methods can help investors achieve stronger performance outcomes.

What is data science?

First things first: What is data science? Data science is a multidisciplinary toolkit focused on uncovering actionable insights from large sets of data. Data scientists, including those at AlphaGeo, use advanced techniques from computer science, predictive analytics, statistics, geospatial analysis and machine learning to organize and parse through massive datasets, including by exploring and connected previously disparate data sources. These help us answer investment questions while generating unobvious insights with new avenues for exploration and analysis.

How can it help?

(1) Information Arbitrage Through Market Intelligence at Scale

How can data science help?

Constructing benchmark-beating portfolios requires a first-mover informational edge about markets and available opportunities. Traditionally, relationships have driven real estate information asymmetry and performance advantages. Now real estate investors can tap on a new lever of information arbitrage: geospatial data science.

Data science arms analysts with the ability to ingest thousands of real-world variables at unprecedented speed and scale, including potentially material information that may be overlooked in conventional analysis. This includes both traditional variables — NOI, GDP, demographics, etc. — and alternative variables that have been made possible by advances in data science, from satellite imagery to sentiment analysis. Well-constructed AI and machine learning algorithms then help parse through these large datasets to identify patterns and signals, offering novel insights or identifying opportunities.

How can investors achieve this?

Data science cannot begin without its key ingredient — vast amounts of deep and diversified data. Investors first need a data lake of well-engineered datasets and variables relevant to market analysis. A sophisticated data lake includes both traditional and alternative data made possible by advances in data science such as such as geospatial data, which can include but not be limited to data with a geolocation footprint as derived from satellite, sensors, transactions, beacons, surveys and more. Most of these datasets and variables, however, must first be “engineered” for use. This is a complex task where data engineers transform raw data into data or analytics relevant and usable to real estate analysts. A data pipeline then stores and processes large amounts of data, structured and unstructured, according to purpose-built algorithms.

How can AlphaGeo help?

Building such a data lake takes time, money, and effort. AlphaGeo offers investors an easy solution with our data lake of 44.6 billion datapoints spanning categories such as real estate, climate, macroeconomics, demographics, infrastructure, and more. We run more than 30 automated pipelines on this wide data lake, allowing us to unravel market complexities and create future scenarios that help investors execute both defensive risk management strategies and offensive acquisition strategies with greater conviction. In addition, we also blend client data into a proprietary data lake to generate more refined insights.

(2) Identify Drivers of Market Performance

How can data science help?

Real estate has traditionally been a heuristic, intuition and relationship driven industry. Investors looking to reduce subjectivity in decision-making can use data science methods to identify empirically proven performance drivers, including less obvious variables that would be missed by traditional analysis or instinct. Moreover, each real estate transaction is unique, and influenced by a complex range of market, socioeconomic, and other factors. Analyzing large numbers of unique transactions to identify key variables that impact performance specific to selected locations and property types would be challenging without data science.

How can investors achieve this?

Data scientists use machine learning algorithms that parse large amounts of data to uncover patterns and correlations to identify variables that matter most to a certain defined outcome –in this case, real estate performance. To ensure that the right pool of variables — known as ‘features’ — are used by these algorithms, data scientists use feature extraction and selection techniques to select features most likely to be relevant, while ignoring irrelevant or redundant features. This process, known as feature engineering, is technically challenging and time-consuming, but key to ensuring that only useful and relevant features are used as input data in any machine learning algorithm.

How can AlphaGeo help?

Given the technical challenges of machine learning driven analysis, AlphaGeo’s platform and advisory services give investors the level of expertise required for feature engineering and model development without having to invest time and money into building in-house teams. All of AlphaGeo’s models have been trained to identify the most influential variables and the degree of their impact on future asset performance under multiple climate change scenarios.

As real estate asset managers face challenging headwinds arising at the confluence of climate, economic and social pressures, let AlphaGeo help turn these forces into tailwinds by investing in resilience. Feel free to contact us and learn more at www.climatealpha.ai.

Originally published January 2024