Nowcast correction function using the quasipoisson distribution.

For more details see the help vignette: vignette("nowcast", package="attrib")

Usage

nowcast_correction_fn_quasipoisson(data, n_week_adjusting, offset, date_0)

Arguments

data

Data generated with nowcast_aggregate containing the part of the dataset that the model should train on.

n_week_adjusting

Number of weeks to correct

offset

Boolian value which is set to true if offset(log(pop)) is a part of the formula

date_0

Date of aggregation.

Value

nowcast_correction_object including corrected data for all weeks in n_wwk_adjust and the model fits for all weeks

Examples

data<- data.table::as.data.table(data_fake_nowcasting_county_aggregated)
data <- data[location_code == "county_nor03"]
n_week_adjusting <- 5
n_week_train <- 52
n_week_start <- n_week_adjusting + n_week_train
date_0 <- data[nrow(data),]$cut_doe #last date in the dataset, assume the dataset is ordered.
data <- data[cut_doe >= (date_0 - n_week_start*7 + 1), ]
nowcast_correction_object <- nowcast_correction_fn_quasipoisson(data,
     n_week_adjusting, offset = "log(pop)", date_0 )