Skip to contents

varpred

The development of this package is motivated by the water, sanitation, and hygiene (WaSH) data in which we were interested in investigating the contribution of demographic and socio-economic factors to improved WaSH indicators among the slum dwellers in Nairobi, Kenya. We noticed that the predictions we generated using the existing packages consistently over- or under- estimated the observed proportions; and did not align well with the observed data points. In other words, what we call . There are several (challenges) reasons for this, including:

  • the choice of the
  • uncertainty estimation – the choice of for computing confidence intervals
  • biases induced by non-linear averaging due to non-linear transformation in generalized linear models

It implements two approaches for constructing outcome plots (prediction and effect plots). These include:

  • mean-based approach
  • observed-value approach

It can also be used to generate bias-corrected prediction and effect estimates for generalized linear models involving non-linear link functions, including models with random effects. This package complements the existing ones by providing:

  • a straightforward way to generate effects plots
  • a robust way to correct for non-linear averaging bias in generalized (mixed) models

Installation

You can install the development version of varpred from GitHub with:

# install.packages("remotes")
remotes::install_github("CYGUBICKO/varpred")

Example

We use mtcars data to show outcome plots:

  • isolate=TRUE to generate effect plot
  • isolate=FALSE to generate prediction plot
library(varpred)
library(ggplot2)

## Set theme for plots
varpredtheme()

## Fit the model
mod <- lm(mpg ~ wt + hp, mtcars)

## Effect
ef <- varpred(mod, "wt", isolate=TRUE, modelname="effect")
plot(ef)

## Prediction
pred <- varpred(mod, "wt", isolate=FALSE, modelname="prediction")
print(plot(pred)
    + scale_color_brewer(palette = "Dark2")
)

## Compare effect and prediction
all_v <- combinevarpred(list(ef, pred))
p1 <- plot(all_v)

## Add observed data
print(p1
    + geom_point(data=mtcars, aes(x=wt, y=mpg), col="grey")
    + labs(colour="Method", linetype="Method")
    + scale_color_brewer(palette = "Dark2")
)