Demo : Parky - Predictive parking
In March 2020, just hours after the lockdown was established in France, our Data Science team at Engie-France realized that we needed to revise our prediction algorithms. A significant shift in behavior was inevitable, and no one knew how it would affect energy consumption.
One of the main challenges we faced was proving that a behavioral change was occurring, even though we only had a few data points from after the lockdown. Initially, plotting daily and monthly consumption data did not reveal any major change. However, through further analysis, we were able to establish a shift in behavior caused by the lockdown. This led to the development of an algorithm capable of detecting and quantifying changes in behavior. Using this measurement, we were able to analyze post-lockdown behavior and identify distinct groups with different consumption patterns.
Later, a study conducted by INSEE [1] showed similar results to what we had observed.
