flatfile {Distance}R Documentation

The flatfile data format

Description

Distance allows loading data as a "flat file" and analyse data (and obtain abundance estimates) straight away, provided that the format of the flat file is correct. One can provide the file as, for example, an Excel spreadsheet using readxl::read_xls in or CSV using read.csv.

Details

Each row of the data table corresponds to one observation and must have a the following columns:

Note that in the simplest case (one area surveyed only once) there is only one Region.Label and a single corresponding Area duplicated for each observation.

The example given below was provided by Eric Rexstad.

Examples

## Not run: 
library(Distance)
# Need to have the readxl package installed from CRAN
require(readxl)

# Need to get the file path first
minke.filepath <- system.file("minke.xlsx", package="Distance")

# Load the Excel file, note that col_names=FALSE and we add column names after
minke <- read_xlsx(minke.filepath, col_names=FALSE)
names(minke) <- c("Region.Label", "Area", "Sample.Label", "Effort",
                  "distance")
# One may want to call edit(minke) or head(minke) at this point
# to examine the data format

## perform an analysis using the exact distances
pooled.exact <- ds(minke, truncation=1.5, key="hr", order=0)
summary(pooled.exact)


## Try a binned analysis
# first define the bins
dist.bins <- c(0,.214, .428,.643,.857,1.071,1.286,1.5)
pooled.binned <- ds(minke, truncation=1.5, cutpoints=dist.bins, key="hr",
                    order=0)

# binned with stratum as a covariate
minke$stratum <- ifelse(minke$Region.Label=="North", "N", "S")
strat.covar.binned <- ds(minke, truncation=1.5, key="hr",
                         formula=~as.factor(stratum), cutpoints=dist.bins)

# Stratified by North/South
full.strat.binned.North <- ds(minke[minke$Region.Label=="North",],
                  truncation=1.5, key="hr", order=0, cutpoints=dist.bins)
full.strat.binned.South <- ds(minke[minke$Region.Label=="South",],
                     truncation=1.5, key="hr", order=0, cutpoints=dist.bins)

## model summaries
model.sel.bin <- data.frame(name=c("Pooled f(0)", "Stratum covariate",
                                   "Full stratification"),
                            aic=c(pooled.binned$ddf$criterion,
                                  strat.covar.binned$ddf$criterion,
                                  full.strat.binned.North$ddf$criterion+
                                  full.strat.binned.South$ddf$criterion))

# Note model with stratum as covariate is most parsimonious
print(model.sel.bin)

## End(Not run)

[Package Distance version 1.0.3 Index]