cuttlefish.model (version 1.0)

delta.glm: LPUE standardisation using Delta-GLM method

Description

The delta.glm function enables the standardization of observed Landings Per Unit Effort (LPUE) collected by commercial fishing vessels using the Delta-GLM methodology. It consists in a combination of a binomial error GLM which explains the presence/absence of the stock and a Gaussian error GLM which explains the abundance of the resource. The standardization is performed using 4 explaining variables, the fishing season, the month, the ICES rectangle and the engine power of the vessel.

Usage

delta.glm(input.data)

Arguments

input.data

Must be filled with an object of class data frame with 4 explaining variables named "fishing.season", "month", "rectangle", "power.class" and one explained variable named "lpue".

Value

binomial.glm

Stores the result of the binomial error GLM

binomial.summary

Stores the summary of the binomial error GLM

binomial.residuals

Stores the residuals of the binomial error GLM

binomial.fit

Stores the fitted values of the binomial error GLM

gaussian.glm

Stores the result of the gaussian error GLM

gaussian.summary

Stores the summary of the gaussian error GLM

gaussian.residuals

Stores the residuals of the gaussian error GLM

gaussian.fit

Stores the fitted values of the gaussian error GLM

predicted.lpue

Stores the standardized LPUE for each quadruplet year, month, rectangle, power.class

Details

The 4 explaining variables ("fishing.season", "month", "rectangle", "power.class") can be either of class integer, character or factor. The explained variable "lpue" must be of class numeric and positive or null. The data frame can include more variables than the 5 mentioned above but they will not be used in the function.

References

Gras, M., Roel, B. A., Coppin, F., Foucher, E. and Robin, J.-P. (2014). A two-stage biomass model to assess the English Channel cuttlefish (Sepia officinalis L.) stock. Submitted to ICES Journal of Marine Science.

See Also

glm

Examples

data(fr.data.lpue)

fr.delta.glm<-delta.glm(input.data=fr.data.lpue)


par(mfrow = c(2,2))
#Histogram of the binomial error GLM residuals
hist(fr.delta.glm$binomial.residuals)

#Plot with the fitted data on the x axis and and the re
plot(fr.delta.glm$binomial.fit, fr.delta.glm$binomial.residuals)

#QQplot of the residuals from the binomial error GLM
qqnorm(fr.delta.glm$binomial.residuals)
qqline(fr.delta.glm$binomial.residuals)


par(mfrow = c(2,2))
#Histogram of the residuals from the Gaussian error GLM
hist(fr.delta.glm$gaussian.residuals)
	
#Plot of fitted values vs residuals from the Gaussian error GLM
plot(fr.delta.glm$gaussian.fit,fr.delta.glm$gaussian.residuals)
qqnorm(fr.delta.glm$gaussian.residuals)
qqline(fr.delta.glm$gaussian.residuals)

#Aggregation of the standardised LPUE per year. Aggregation 
#can be done on the 3 other factors in the same way.
fr.yearly.lpue<-aggregate(fr.delta.glm$predicted.lpue$st.lpue, 
list(fr.delta.glm$predicted.lpue$fishing.season), FUN="mean")
fr.yearly.lpue<-data.frame(c(1900:1905), fr.yearly.lpue)
colnames(fr.yearly.lpue)<-c("year","fishing.season","fr.st.lpue")