Paper lead author here - yes, usage is multi-factorial and there are hidden factors we may not know about. You can see the Paris August holidays in our visualiz

Paper lead author here - yes, usage is multi-factorial and there are hidden factors we may not know about. You can see the Paris August holidays in our visualizations of the models, and I was recently told about Goteborg July holidays. So the model with four variables (UTCI, rain, hour of day, day of year) is best, but then that is hard to visualize.