fdbk_dt_conttable: Calculates stratified contingency table entries (above...

In rfxf/Rfdbk: Handling NetCDF feedback files

Description

Calculates stratified contingency table entries (above threshold) for a data table

Usage

fdbk_dt_conttable(DT, vars = NULL, thrs = NULL, by = NULL, cores = 1)

Arguments

DT	data.table with relevant information
vars	character vector of varnos (if NULL take from DT)
thrs	list of vectors of thresholds for each varno (if NULL threshold are generated from quantiles)
by	stratify contingency entries by these DT columns
cores	number of CPU cores to split the calculation (helps for larger data tables)

Value

data.table with columns varno,thr, hits,false,miss,corrneg and the arguments of 'by'

Author(s)

Felix <[e-mail]>

Examples

#EXAMPLE (CSI for quantile thresholds) 
require(ggplot2)
fnames                = system("ls ~/examplesRfdbk/*/synop/verSYNOP.*",intern=T)
cond                  = list(veri_description="grepl('forecast',veri_description)",
                             veri_forecast_time="veri_forecast_time%in%c(2400,4800,7200,9600,12000,14400,16800)")
columnnames           = c("obs","veri_data","varno","veri_model","veri_forecast_time","statid","veri_initial_date")
DT                    = fdbk_dt_multi_large(fnames,cond,columnnames,20)
vars                  = c('1','3','4','29')
thrs                  = list('1'=c(50,60),'3'=c(-5,0,5),'4'=c(-5,0,5),'29'=c(.4,.6,.8))
xx                    = fdbk_dt_conttable(DT,vars=vars,thrs=thrs,by=c("veri_model","veri_forecast_time","varno"),cores=10)
CSI                   = xx[,list(csi =(hit)/(hit + miss + false) ),by=c("veri_forecast_time","veri_model","thr","varno")]
CSI[,varno:=varno_to_name(varno,T)]
ggplot(CSI,aes(x=thr,y=csi,color=factor(veri_forecast_time),linetype=factor(veri_model),group=interaction(varno,veri_forecast_time,veri_model)))+
geom_line()+
ggtitle("CSI")+
facet_wrap(~varno,scales="free_x")

rfxf/Rfdbk documentation built on May 27, 2019, 7:22 a.m.