You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Via the grid_type argument to haldensify(), which is passed to format_long_hazards(), discretization and binning of the outcome variable (A) in the conditional density estimation routine g(A|W) is handled by functions that bin points based on an equal partitioning along the range of A ("equal_range") or into possibly uneven bins that contain roughly the same number of observations each ("equal_mass"), as happens at https://github.com/nhejazi/haldensify/blob/master/R/utils.R#L51-L63. It would be beneficial in many applications to allow the user to specify the binning/discretization themselves, given that these two simple routines may lead to suboptimal performance in many settings, e.g., when there are rare instances that fall far away from others among the set of observations. It would be beneficial to allow the relevant argument to take as input a set of cut points along the observed support at which the variable A will be discretized.
The text was updated successfully, but these errors were encountered:
Via the
grid_type
argument tohaldensify()
, which is passed toformat_long_hazards()
, discretization and binning of the outcome variable (A) in the conditional density estimation routine g(A|W) is handled by functions that bin points based on an equal partitioning along the range of A ("equal_range"
) or into possibly uneven bins that contain roughly the same number of observations each ("equal_mass
"), as happens at https://github.com/nhejazi/haldensify/blob/master/R/utils.R#L51-L63. It would be beneficial in many applications to allow the user to specify the binning/discretization themselves, given that these two simple routines may lead to suboptimal performance in many settings, e.g., when there are rare instances that fall far away from others among the set of observations. It would be beneficial to allow the relevant argument to take as input a set of cut points along the observed support at which the variable A will be discretized.The text was updated successfully, but these errors were encountered: