prm: Modified panel regression model
In datalimited/datalimited: Stock Assessment Methods for Data-limited Fisheries

Description Usage Arguments Value References Examples

Format for, fit, and predict from a panel regression model. The datalimited package includes cached data that can be used with fit_prm and a cached model that can be used with predict_prm. These functions can also be used with new data.

format_prm: Format a series of predictors for a panel regression

fit_prm: Fit a panel regression

predict_prm: Predict from a panel regression model

format_prm(year, catch, species_cat, bbmsy = NULL)

fit_prm(dat, eqn = log(bbmsy) ~ max_catch + mean_scaled_catch + scaled_catch +
  scaled_catch1 + scaled_catch2 + scaled_catch3 + scaled_catch4 + species_cat +
  catch_to_rolling_max + time_to_max + years_back + initial_slope - 1,
  type = c("lm", "gbm"), ...)

predict_prm(newdata, model = datalimited::ram_prm_model, ci = FALSE)

`year`	A numeric vector of years
`catch`	A numeric vector of catches
`species_cat`	A single character value of a species category. Can be from any set of species categories as long as the same set are used in model fitting and extrapolation.
`bbmsy`	A numeric vector of B/B_MSY (only needed if passing the output to `fit_prm`; leave as `NULL` if passing output to `predict_prm`)
`dat`	A data frame created by `format_prm`
`eqn`	A formula describing the regression
`type`	Either linear model (`"lm"`; the original method) or a `gbm` model (`"gbm"`)
`...`	Anything extra to pass to `gbm`
`newdata`	A data frame to predict on that has been formatted with `format_prm`.
`model`	A linear model to predict from built from `fit_prm`. Defaults to a cached version of a model fit to the RAM database, `ram_prm_model`.
`ci`	Should confidence intervals on B/B_MSY be returned?

format_prm: A data frame formatted for use with fit_prm or predict_prm

fit_prm: A linear model for use with predict_prm

predict_prm: If ci = FALSE, a vector of predictions of B/B_MSY. If ci = TRUE, a data frame. Note that the model is fitted to log(B/B_MSY) and the output from predict_prm is exponentiated so the prediction represents an estimate of median B/B_MSY.

Costello, C., D. Ovando, R. Hilborn, S. D. Gaines, O. Deschenes, and S. E. Lester. 2012. Status and Solutions for the World's Unassessed Fisheries. Science 338:517-520.

# combine two built in datasets:
# d <- dplyr::inner_join(ram_ts, spp_categories, by = "scientificname")

# ram_prm_dat is built into the package; it can be created with:
# ram_prm_dat <- plyr::ddply(d, "stockid", function(x) {
#   format_prm(year = x$year, catch = x$catch, bbmsy = x$bbmsy_ram,
#     species_cat = x$spp_category[1L])
# })

# ram_prm_model is built into the package, it is created in the same manner
# as this:
# m <- fit_prm(ram_prm_dat)

# now predict B/Bmsy:
d <- subset(ram_prm_dat, stockid == "BGRDRSE")
x <- predict_prm(d) # use the built in model
plot(x)

# with confidence intervals:
library("ggplot2")
x <- predict_prm(d, ci = TRUE)
ggplot(x, aes(year, bbmsy_q50)) +
  geom_ribbon(aes(ymin = bbmsy_q2.5, ymax = bbmsy_q97.5), fill = "#00000040") +
  geom_line() +
  ylab(expression(B/B[MSY]))

# format and predict on a new fake dataset:
set.seed(1)
dat <- format_prm(
  year = 1:10,
  catch = rlnorm(10),
  species_cat = "Cods, hakes, haddocks")
head(dat)
predict_prm(dat)