Using Data Science to Support Real Estate Investment Strategy 2021 | Global Risk Institute in Financial Services | 27 October 2021 | Virtual Event

Data science has applications across the whole real estate investment stack. It certainly has application in the financial side of things, such as finding alpha and being able to more efficiently understand where there is investment opportunity. Please join Erkan Yonder, Laurentian Bank Professor of Real Estate and Finance at Concordia University - John Molson School of Business as he discusses his paper, Neighbourhood Effects, Immigration and Real Estate Valuation: A Machine Learning Approach.

Important takeaways:

- This project uses a machine learning model (LASSO) to determine the best neighborhood predictors.

• Among 105 neighborhood variables, the LASSO machine learning model detects six local immigration variables, which is important for the Canadian real estate markets.
• A higher share of external immigrants, non-permanent residents, and Arab and South Asian population in a neighborhood are shown to increase house prices.
• The model also reflects other important neighborhood effects such as local income or unemployment.
- This project also develops a benchmark score, which can help investors better assess their assets and investments using neighborhood factors. Controlling for the benchmark score increases in-sample Adjusted R-Squared from 49% to 70%, that is only one variable instead of hundreds of zip code fixed effects.

- The machine learning model framework improves prediction accuracy up to 30% compared to relative models.

2021 Speaker

Erkan Yonder
Laurentian Bank Professor of Real Estate and Finance at Concordia University - John Molson School of Business