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.
- 📊 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 Coxpspline() - 📉 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
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")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)# 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()`).
# 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()`).
# 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()`).
# Default: dotted lines splineplot(fit_gam_cox, dat, ribbon_ci = FALSE) # Alternative: ribbon/shaded area splineplot(fit_gam_cox, dat, ribbon_ci = TRUE)# 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)
# 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
| Model Type | Model Function | Spline Types | Outcome |
|---|---|---|---|
| GAM | mgcv::gam() | s(), te(), ti() | HR, OR, RR, Effect |
| GLM | stats::glm() | ns(), bs() | OR, RR, Effect |
| Linear | stats::lm() | ns(), bs() | Effect |
| Cox | survival::coxph() | ns(), bs(), pspline()* | HR |
*Note: pspline() has limited support due to its internal structure. We recommend using ns() or bs() with Cox models for optimal results.
| Parameter | Description | Default |
|---|---|---|
fit | Fitted model object | Required |
data | Data frame used for fitting | Required |
xvar | Variable name for x-axis | Auto-detected |
by_var | Interaction variable | Auto-detected |
refx | Reference x value | Median of x |
xlim | X-axis limits | Data range |
ylim | Y-axis limits | Auto |
show_hist | Show histogram | TRUE |
ribbon_ci | Use ribbon CI style | FALSE |
log_scale | Use log scale for y-axis | FALSE |
show_ref_point | Show reference point marker | TRUE |
xlab | X-axis label | Variable name |
ylab | Y-axis label | Auto by model |
ylab_right | Right y-axis label | “Percent of Population” |
If you use splineplot in your research, please cite:
citation("splineplot")- Documentation: Visit our pkgdown site
- Bug reports: GitHub Issues
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.
Apache License 2.0 © Jinseob Kim / Zarathu
Special thanks to the developers of mgcv, survival, and ggplot2 packages.