4.8 Linear regression with count data - negative values
The results of fitting a least squares regression model to the Bikeshare
data provides some reasonable results:
as weather progressively worsens, the number of bikers decreases (coefficients become negative wrt baseline)
the coefficients associated with season and time of day match expected patterns (lowest in winter, and highest during peak commute times)
Problem 1: model predicts negative numbers of bikers at times