sandeshregmi

#' Simulator for projecting bed occupancy
#'
#' This function predits bed occupancy from admission data (dates, and numbers
#' of admissions on these days). Duration of hospitalisation is provided by a
#' function returning `integer` values for the number of days in hospital.
#' 
#' @param dates A vector of dates, ideally as `Date` but `integer` should work too.
#'
#' @param n_admissions An `integer` vector giving the number of admissions
#'   predicted for each date in `dates`.

sandeshregmi

#' Project bed occupancy from admissions
#'
#' This projects bed occupancy using admission incidence stored as a
#' `projections` object, and a distribution of length of stay (los). This is a
#' wrapper around `simulate_occupancy`, which essentially applies this function
#' to different admission trajectories and collects outputs into a single
#' `projections` object.
#'
#' @param x a `projections` object storing forecast of daily admissions

sandeshregmi

sandeshregmi

sandeshregmi

sandeshregmi

def pop_diff(popA, popB, bin_boundaries=None, num_bins=10):
    """
    Compute difference between two populations using the PSI / IV formula
    
    $$\Sigma_{i} (p_i^B - p_i^a)*\ln(\frac{p_i^B}{p_i^A})$$
    
    Note:
        Counts missing values in a separate bin to test for information and shift.
        Raises ValueError if popA has no nans but popB does.
    

sandeshregmi

sandeshregmi

def f1_loss(y_true:torch.Tensor, y_pred:torch.Tensor, is_training=False) -> torch.Tensor:
    '''Calculate F1 score. Can work with gpu tensors
    
    The original implmentation is written by Michal Haltuf on Kaggle.
    
    Returns
    -------
    torch.Tensor
        `ndim` == 1. 0 <= val <= 1
    

sandeshregmi