Nicolas Martinez - Statistics Canada, Talk: Earth Observation Data & Artificial Intelligence to Estimate Construction Starts

Investment in building construction statistics for the residential and non-residential sectors account for approximately 5% of GDP and are used extensively by internal and external clients. The statistics available for the residential and non-residential building construction sectors provide an invaluable portrait of this activity in Canada at the national, provincial and census metropolitan area (CMA) level.

In order to improve the quality of our estimates, while also modernizing data collection, Statistics Canada in partnership with CMHC is launching a pilot project to evaluate the feasibility of using geospatial data in conjunction with machine learning/artificial intelligence to estimate the start of construction for all building types. The project will be developed using a phased approach, starting with very high resolution images to determine the start of construction for select geographies, with subsequent phases potentially detecting the completion of construction. Outputs from the program will be validated using CMHC’s starts and completions data for the residential sector, and field validation will be carried out