In brouwern/avesdemazan: What the Package Does (One Line, Title Case)

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(avesdemazan)

Data

Get the best model

data(best_randslopes)

Coefficinets table - untransformed

Get coefficients and make table

best_fit_tab <- as.data.frame(summary(best_randslopes)$coefficients$cond)

rownames(best_fit_tab) <- c("Intercept", "Location (MAIN)", "Location (MASE)", "Time")

kable(best_fit_tab, caption = "Model output on original (log) scale")

Coefficinets table - transformed?

best_fit_tab <- as.data.frame(summary(best_randslopes)$coefficients$cond)
rownames(best_fit_tab) <- c("Intercept", "Location (MAIN)", "Location (MASE)", "Time")
best_fit_tab2 <- best_fit_tab

# Point estimates
best_fit_tab2$EstimateExp <- NA

best_fit_tab2$EstimateExp[1] <- best_fit_tab2$Estimate[1]
best_fit_tab2$EstimateExp[2] <- best_fit_tab2$Estimate[1] + best_fit_tab2$Estimate[2]
best_fit_tab2$EstimateExp[3] <- best_fit_tab2$Estimate[1] + best_fit_tab2$Estimate[3]

# SE estimates

### GET vcov matrix
vcov <- matrix(unlist(vcov(best_randslopes)), nrow = 4)

best_fit_tab2$SE <- NA

best_fit_tab2$SE[1] <- sqrt(vcov[1,1])
best_fit_tab2$SE[2] <- sqrt(vcov[1,1] + vcov[2,2] + 2*vcov[1,2])
best_fit_tab2$SE[3] <- sqrt(vcov[1,1] + vcov[3,3] + 2*vcov[1,3])

# Confidence intervals
## Lower bound & upper bound
best_fit_tab2$LB <- round(exp(best_fit_tab2$EstimateExp - 1.96*best_fit_tab2$SE), 6)*1000
best_fit_tab2$UB <- round(exp(best_fit_tab2$EstimateExp + 1.96*best_fit_tab2$SE), 6)*1000

## Calculate confidence interval
best_fit_tab2$CI <- paste0("(", best_fit_tab2$LB, ", ", best_fit_tab2$UB, ")")

# Point Estimate
best_fit_tab2$EstimateExp <- round(exp(best_fit_tab2$EstimateExp), 6) * 1000

# Drop time effect row
best_fit_tab2 <- best_fit_tab2[-4,]

# Select point estimate and 95% CI columns
best_fit_tab2 <- best_fit_tab2[,c(5,9)]

# Format
rownames(best_fit_tab2) <- c("Secondary forest (LLAV)", "Introduced forest (MAIN)", "Primary forest (MASE)")
colnames(best_fit_tab2) <- c("Baseline captures per 1,000 net hours", "95% Confidence Interval")
kable(best_fit_tab2, caption = "Transformed model estimates")

table produced above is just intercept terms for main effects - not needed (?)

We estimate that the baseline (time = 0) capture rate per 1,000 net hours is 7.191 (95\% CI (5.383, 9.606)) in secondary forest, 5.820 (95\% CI (4.226, 8.015)) in introduced forest, and 4.423 (95\% CI (3.153, 6.206)) in primary forest. We estimate a 2.8\% (95\% CI (0.9\%, 4.6\%)) decrease in capture rate between years in all habitats. **

brouwern/avesdemazan documentation built on July 26, 2022, 8:38 p.m.