Visualization of Spline Effects in GAM and GLM Models.

splineplot

The splineplot package provides a unified interface for visualizing spline effects from GAM (Generalized Additive Models) and GLM (Generalized Linear Models) in R. It creates publication-ready plots with confidence intervals, supporting various model types including Linear, Logistic, Poisson, and Cox proportional hazards models.

Key Features

• 📊 Unified visualization for spline effects across different model types
• 🎨 Publication-ready plots with customizable aesthetics
• 📈 Automatic detection of spline terms and interaction variables
• 🔄 Support for multiple spline types: GAM smooths (s(), te(), ti()), GLM splines (ns(), bs()), and Cox pspline()
• 📉 Flexible confidence intervals: dotted lines or ribbon style
• 📊 Built-in histograms showing data distribution
• 🎯 Reference point marking with automatic SE = 0 at reference value
• 🔀 Interaction term support with by-variable visualization

Installation

You can install the released version of splineplot from CRAN:

install.packages("splineplot")

Or install the development version from GitHub:

# install.packages("devtools")
devtools::install_github("jinseob2kim/splineplot")

Basic Usage

library(splineplot)
library(mgcv)
library(survival)
library(splines)
library(ggplot2)

# Generate sample data
set.seed(123)
n <- 500
x <- rnorm(n, mean = 35, sd = 8)
lp <- -0.06*(x - 35) + 0.0009*(x - 35)^3/(8^2)
time <- rexp(n, rate = exp(lp))
status <- rbinom(n, 1, 0.8)
binary_y <- rbinom(n, 1, plogis(lp))

dat <- data.frame(x, time, status, binary_y)

GAM with Cox family

# Fit GAM Cox model
fit_gam_cox <- gam(time ~ s(x),
                   family = cox.ph(), weights = status, data = dat)

# Create spline plot
splineplot(fit_gam_cox, dat,
          ylim = c(0.2, 2.0),
          xlab = "Age (years)",
          ylab = "Hazard Ratio")
#> Using 'x' as x variable
#> Using refx = 35.17 (median of x)
#> Warning: Removed 67 rows containing missing values or values outside the scale range
#> (`geom_line()`).
#> Warning: Removed 56 rows containing missing values or values outside the scale range
#> (`geom_line()`).
#> Warning: Removed 62 rows containing missing values or values outside the scale range
#> (`geom_line()`).

Logistic regression with natural splines

# Fit logistic model with natural splines
fit_glm <- glm(binary_y ~ ns(x, df = 4),
               family = binomial(), data = dat)

# Create spline plot
splineplot(fit_glm, dat,
          ylim = c(0.2, 2.0),
          ylab = "Odds Ratio")
#> Using 'x' as x variable
#> Using refx = 35.17 (median of x)
#> Warning: Removed 37 rows containing missing values or values outside the scale range
#> (`geom_line()`).
#> Warning: Removed 85 rows containing missing values or values outside the scale range
#> (`geom_line()`).
#> Warning: Removed 59 rows containing missing values or values outside the scale range
#> (`geom_line()`).

GAM with interaction terms

# Add a grouping variable
dat$group <- factor(sample(c("A", "B"), n, replace = TRUE))

# Fit model with interaction
fit_interaction <- gam(time ~ s(x, by = group),
                      family = cox.ph(),
                      weights = status,
                      data = dat)

# Plot with interaction
splineplot(fit_interaction, dat,
          ylim = c(0.2, 2.0))
#> Using 'x' as x variable
#> Detected interaction with 'group'
#> Using refx = 35.17 (median of x)
#> Warning: No shared levels found between `names(values)` of the manual scale and the
#> data's fill values.
#> Warning: Removed 136 rows containing missing values or values outside the scale range
#> (`geom_line()`).
#> Warning: Removed 116 rows containing missing values or values outside the scale range
#> (`geom_line()`).
#> Warning: Removed 126 rows containing missing values or values outside the scale range
#> (`geom_line()`).

Advanced Features

Customizing confidence intervals

# Default: dotted lines
splineplot(fit_gam_cox, dat, ribbon_ci = FALSE)

# Alternative: ribbon/shaded area
splineplot(fit_gam_cox, dat, ribbon_ci = TRUE)

Log scale for odds/rate/hazard ratios

# Use log scale for y-axis
splineplot(fit_glm, dat, log_scale = TRUE)
#> Using 'x' as x variable
#> Using refx = 35.17 (median of x)

Custom reference values

# Set custom reference point (default is median)
splineplot(fit_gam_cox, dat,
          refx = 40,  # Reference at x = 40
          show_ref_point = TRUE)  # Show diamond marker
#> Using 'x' as x variable

Supported Model Types

*Note: pspline() has limited support due to its internal structure. We recommend using ns() or bs() with Cox models for optimal results.

Function Parameters

Citation

If you use splineplot in your research, please cite:

citation("splineplot")

Getting Help

• Documentation: Visit our pkgdown site
• Bug reports: GitHub Issues

Contributing

Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.

License

Apache License 2.0 © Jinseob Kim / Zarathu

Acknowledgments

Special thanks to the developers of mgcv, survival, and ggplot2 packages.