retfit: Fit linear regression with covariate value lagged

In eeholmes/SardineForecast: Indian Oil Sardine Data, Code and References for Papers on Forecasting

Description

Fit a linear regression model where the covariate value used as explanatory variable is lagged by a user specified number of quarters. The effect of a covariate can be specified to be quarter specific, that is catch in Q1 may respond differently to the covariate than catch in Q4. Note: this is an outdated function not used since I switched to GAMs.

Usage

retfit(
  covname,
  catch.qtr = 1:4,
  lag = 0,
  boxes = 1:14,
  qtr.effect = FALSE,
  startyear = 1956,
  endyear = 2016
)

Arguments

covname	name of the covariate. Must be a colname in seio_covariates_qtr. If one of the covariates is from the boxes, then specify the part of the name in front of the number.
catch.qtr	what quarter(s) of catch to use as the response
lag	the lag specified in number of quarters. 0 means no lag.
boxes	for covariate from a box, the box number or numbers to use
qtr.effect	whether to make the covariate effect on catch depend on the quarter in which the catch is from
startyear	first year
endyear	last year

Value

An lm object of the fit

Examples

#Effect of precip in box 5 in qtr 1 in the previous year on catch
retfit("Precip", lag=1, boxes=5)

eeholmes/SardineForecast documentation built on July 17, 2021, 2:56 a.m.